Long-term changes in health-related quality of life among Australian adults with psychological distress: a 16-year perspective.

To identify the characteristics of the variables, frequency and descriptive statistical analysis were performed. Independent t-test and chi-square test were conducted to check differences among continuous and categorical variables respectively. Table  2 reports the general and clinical characteristics of the respondents. The mean age of the respondents was 44.88 years, and the most common age group was 24–44 years (34%). The percentage of females was recorded at 53%. The sample comprised 96% non-Indigenous participants, English speaking (89%), living in major cities (63%), married (62%), employed (64%), obese (36%), not active club member (63%), non-smokers (82%), occasional drinkers (42%), had 12 years or below level of education (44%), had high level of physical activity (35%) and had annual disposable income <$30,000 (43%). Approximately, on average, participants with PD were six years younger than those without PD. Psychological distress was more prevalent among females (55%), in age group 24–44 years (38%), participants having low level of physical activity (37%), and obese (44%). Table 2 Sociodemographic and clinical characteristics of study respondents across all waves from the Household, income and labour dynamics in Australia survey (2007–2021) [no psychological distress vs. psychological distress] Total number of (observation) Respondents with no psychological distress (K10: 10–19) Respondents with psychological distress (K10: 20–50) P -Value Characteristics ( n  = 130,388) ( n  = 88,843) ( n  = 41,545) Age  Average in years ( n ) 44.881 46.850 40.694 < 0.01 Sex  Male % ( n ) 47 (61,682) 48 (42,865) 45 (18,817)  Female % ( n ) 53 (68,706) 52 (45,978) 55 (22,728) < 0.01 Age group  15–24% ( n ) 17 (22,775) 14 (12,865) 24 (9,910)  24–44% ( n ) 34 (44,421) 32 (28,684) 38 (15,737)  45–64% ( n ) 31 (39,769) 33 (29,147) 26 (10,622)  65+%( n ) 18 (23,423) 20 (18,147) 13 (5,276)  0.01 Speak language other than English  Yes %( n ) 11 (12,002) 9 (7,134) 14 (4,868)  No %( n ) 89 (100,762) 91 (69,823) 86 (30,939) < 0.01 Remoteness area  Major cities % ( n ) 63 (80,571) 63 (55,240) 62 (25,331)  Inner regional Australia % ( n ) 26 (33,022) 26 (22,465) 26 (10,557)  Outer Regional, Remote and Very Remote Australia % ( n ) 11 (14,568) 11 (9,682) 11 (4,886) > 0.01 Marital Status  Legally married & De Facto % ( n ) 62 (81,352) 67 (59,670) 52 (21,682)  Separated, Divorced & Widowed % ( n ) 13 (17,574) 13 (11,386) 15 (6,188)  Never married and not de facto % ( n ) 24 (31,441) 20 (17,780) 33 (13,661) Education level  Year 12 and below % ( n ) 44 (57,406) 41 (36,285) 51 (21,121)  Certificate III/IV % ( n ) 22 (28,633) 22 (19,303) 22 (9,330)  Bachelor/honours & diploma/adv diploma % ( n ) 23 (30,435) 25 (22,471) 19 (7,964)  Postgrad-master or doctorate, grad diploma & grad certificate %( n ) 11 (13,839) 12 (10,745) 7 (3,094) < 0.01 Current labour force status  Employed %( n ) 64 (71,800) 66 (50,754) 59 (21,046)  Unemployed %( n ) 4 (4,255) 3 (2,017) 06 (2,238)  Not in labour force %( n ) 33 (36,728) 31 (24,191) 35 (12,537) < 0.01 Physical activity Level  Low %( n ) 32 (11,165) 30 (7,199) 37 (3,966)  Moderate %( n ) 33 (11,475) 34 (8,244) 30 (3,231)  High %( n ) 35 (12,027) 36 (8,604) 32 (3,423) < 0.01 Body Mass Index  Normal %( n ) 35 (39,931) 36 (31,000) 32 (8,931)  Overweight%( n ) 29 (33,031) 30 (26,274) 24 (6,757)  Obesity %( n ) 36 (41,466) 34 (28,925) 44 (12,541) < 0.01 Club membership  Yes %( n ) 37 (42,800) 40 (35,408) 27 (7,392)  No %( n ) 63 (72,729) 60 (52,737) 73 (19,992) < 0.01 Smoking status  No %( n ) 82 (95,091) 85 (75,107) 73 (19,984)  Yes %( n ) 18 (20,420) 15 (13,023) 27 (7,397) 5 days per week) %( n ) 14 (14,277) 15 (11,675) 11 (2,602)  Regular: 2–4 days per week %( n ) 25 (24,755) 26 (19,528) 22 (5,227)  Occasional: <2 days per week %( n ) 42 (41,927) 41 (31,681) 43 (10,246) < 0.01 Annual disposable income  $1-$30,000%( n ) 43 (56,682) 40 (35,661) 51 (21,021)  $30,001-$60,000%( n ) 33 (42,705) 33 (29,650) 31 (13,055)  $60,001-$100,000%( n ) 17 (22,354) 19 (16,756) 13 (5,598)  $100,001-$200,000%( n ) 6 (7,183) 6 (5,636) 4 (1,547)  >$20,0001%( n ) 1 (1,464) 1 (1,140) 1 (324) > 0.01 Indigenous status, speak language other than English, remoteness area, marital status, education level, current labour force status, physical activity level, body mass index, club membership, smoking status, alcohol consumption was not available for 41,155, 17,624, 2,227, 21, 75, 54,333, 95,721, 15,960, 14,859, 14,877, 30,234, respondents respectively K10 Kessler Psychological Distress Scale * T-test and Chi square tests were applied for continuous and categorical variables respectively. Sociodemographic and clinical characteristics of study respondents across all waves from the Household, income and labour dynamics in Australia survey (2007–2021) [no psychological distress vs. psychological distress] Indigenous status, speak language other than English, remoteness area, marital status, education level, current labour force status, physical activity level, body mass index, club membership, smoking status, alcohol consumption was not available for 41,155, 17,624, 2,227, 21, 75, 54,333, 95,721, 15,960, 14,859, 14,877, 30,234, respondents respectively K10 Kessler Psychological Distress Scale * T-test and Chi square tests were applied for continuous and categorical variables respectively. Comparison of respondents having PD with no PD shows that the two groups were different in sex ( p  < 0.01). The proportion of respondents having active club membership (27%: PD and 40%: no PD) significantly differed between the two groups ( p  < 0.01). Similarly, the proportion of smokers (27%: PD and 15%: no PD) also significantly differed between the two groups. The age distribution of PD and no PD cohort was statistically different from each other ( p  < 0.01). There were significant differences in the distribution of marital status ( p  < 0.01), English speaking ( p  < 0.01), education level ( p  < 0.01), employment status ( p  < 0.01), physical activity level ( p  < 0.01), BMI level ( p  < 0.01), and alcohol consumption ( p   0.01), Indigenous status ( p  > 0.01) and level of income ( p  > 0.01). Our simple model (see Additional file Table S1) shows the significant role of time when it is considered as a sole determinant of HRQoL. Time coefficients remain statistically and clinically significant after the inclusion of PD as determinant in the model (see Additional file Table S2). Both tables show that HRQoL declined over time. Table  3 reports the result of the full model with all determinants. Results show that different classifications of PD and HRQoL are negatively associated. The negative gradient of PD steepens with the severity of the disease (from − 0.086 mild PD to -0.177 in severe PD). A similar trend exists regarding association between different classifications of PD and four domains of HRQoL. However, the size of the effect is disproportionate with the highest effect recorded in the domain of mental health (-0.364) in the category of severe PD. The lowest effect of severe PD was recorded in the domain of PF (-0.114). Table 3 Linear mixed effects models with predictors of health-related quality of life and its domains among Australian adults: full sample analysis from the Household, income and labour dynamics in Australia survey (2007–2021) Health State Utilities a Physical Function Role Physical Mental Health Role Emotional Number of observations 111,968 115,318 115,089 116,002 114,976 Psychological distress b  No PD c (K10: 10–19) Ref Ref Ref Ref Ref  Mild PD (K10: 20–29) -0.086 ** -0.049 ** -0.122** -0.156 ** -0.241**  Moderate PD (K10: 25–29) -0.128 ** -0.075 ** -0.176** -0.245 ** -0.391**  Severe PD (K10: 30–50) -0.177 ** -0.114 ** -0.257** -0.364 ** -0.525**  Missing/Undetermined -0.050 ** -0.050 ** -0.097** -0.088 ** -0.161** Age  15–24 years Ref Ref Ref Ref Ref  25–44 years -0.016 ** -0.028 ** -0.064** -0.015 ** -0.051**  45–64 years -0.039 ** -0.099 ** -0.149** -0.009 ** -0.076**  65 + years -0.059 ** -0.183 ** -0.270** 0.007 ** -0.122** Sex  Male Ref Ref Ref Ref Ref  Female -0.012 ** -0.006 ** -0.007** -0.012 ** -0.016** Indigenous Status  Not of indigenous origin Ref Ref Ref Ref Ref  Aboriginal or/and Torres Strait Islander -0.003 -0.022 ** 0.009 -0.000 0.025  Missing/Undetermined 0.002 -0.005 0.013 ** 0.003 0.011** Speak language other than English  Yes Ref Ref Ref Ref Ref  No 0.002 0.005 -0.006 0.002 -0.003  Missing/Undetermined 0.015 0.006 0.009 0.049 -0.089 Remoteness area  Major cities Ref Ref Ref Ref Ref  Inner regional Australia 0.000 -0.002 -0.007* 0.005** 0.008**  Outer Regional, Remote and Very Remote Australia -0.002 * -0.010** -0.021** 0.005** 0.001  Missing/Undetermined -0.002 0.029 0.002 -0.016 0.109 Marital Status  Legally married & De Facto Ref Ref Ref Ref Ref  Separated, Divorced & Widowed -0.018 ** -0.052 ** -0.056** -0.012 ** -0.054**  Never married and not de facto 0.003 * 0.012 ** 0.014** -0.010 ** -0.020**  Missing/Undetermined -0.030 0.053 -0.158 -0.017 -0.011 Education level  Year 12 and below Ref Ref Ref Ref Ref  Certificate III/IV -0.002 0.011** -0.008 * 0.002 -0.004  Bachelor/honours & diploma/adv diploma 0.003 ** 0.030** 0.008 * 0.001 -0.004  Postgrad-master or doctorate, grad diploma & grad certificate 0.007 ** 0.041** 0.015 ** -0.002 0.001  Missing/Undetermined 0.009 0.042** 0.163 ** 0.030 0.149 * Current labour force status  Employed Ref Ref Ref Ref Ref  Unemployed -0.004 * 0.001 -0.003 -0.008 ** -0.019**  Not in labour force -0.001 ** -0.040 -0.082** -0.005 ** -0.046** Physical activity Level  Low Ref Ref Ref Ref Ref  Moderate 0.013 ** 0.028** 0.057** 0.005 ** 0.021**  High 0.021 ** 0.032** 0.073** 0.012 ** 0.030**  Missing/Undetermined 0.020 ** 0.045 ** 0.058** 0.006 0.027 Body Mass Index  Normal Ref Ref Ref Ref Ref  Overweight -0.004 ** -0.011 ** -0.010** 0.001 -0.000  Obesity -0.015 ** -0.038 ** -0.039** -0.003 -0.008**  Missing/Undetermined -0.009 ** -0.028** -0.014** 0.001 0.005 Club membership  Yes Ref Ref Ref Ref Ref  No -0.008 ** -0.017 ** -0.012** -0.012 ** -0.010**  Missing/Undetermined -0.014 ** -0.025 ** -0.019* -0.016 ** -0.006 Smoking status  No Ref Ref Ref Ref Ref  Yes -0.011 ** -0.010 ** -0.008* -0.008 ** -0.022**  Missing/Undetermined -0.013 ** -0.021 ** -0.027* -0.010 ** -0.044** Alcohol Consumption  Abstinent Ref Ref Ref Ref Ref  Daily or Almost Daily: >5 days per week) 0.008 ** 0.030 ** 0.050** -0.001 0.011 **  Regular: 2–4 days per week 0.007 ** 0.034 ** 0.034** -0.002 -0.001  Occasional: <2 days per week 0.006 ** 0.025 ** 0.025** -0.001 0.002  Missing/Undetermined -0.001 0.028 ** 0.021** -0.001 0.018 Annual disposable income  $1-$30,000 Ref Ref Ref Ref Ref  $30,001-$60,000 0.009 ** 0.020 ** 0.032** 0.006 ** 0.024**  $60,001-$100,000 0.012 ** 0.032 ** 0.047** 0.006 ** 0.026**  $100,001-$200,000 0.013 ** 0.035 ** 0.058** 0.007 ** 0.030**  >$20,0001 0.011 ** 0.023 ** 0.044** 0.012 ** 0.032** Year  2007 Ref  2009 0.001 0.001 0.000 0.001 -0.003  2011 -0.003** -0.006 -0.007 * -0.001 -0.006  2013 0.005 0.016 0.006 0.000 0.007  2015 -0.012 0.001 -0.031 -0.052 0.042  2017 0.002 0.014 0.001 -0.002 -0.001  2019 -0.008** -0.008** -0.014** -0.006** -0.020**  2021 -0.007** -0.002 -0.013** -0.013** -0.040** Constant 0.808 ** 0.877 ** 0.902** 0.0804 ** 0.976** a The regression model includes the grouping variable (wave_count1) that distinguishes consistent and inconsistent participants in model as a random effect to account for the variability between consistent and inconsistent groups, assuming that the individuals in the “consistent and inconsistent” groups are distinct and their HRQoL variation is unique to each person b K10 Kessler Psychological Distress scale (K10). The cut off points of K10 are based on the 2001 Victorian Population Health Survey. The cut-off scores were based on how practitioners use the K10 as a screening tool ( https://www.abs.gov.au/ausstats/[email protected]/ProductsbyReleaseDate/4D5BD324FE8B415FCA2579D500161D57 ) c PD Psychological Distress *, ** show P  < 0.05 and 0.01 respectively Linear mixed effects models with predictors of health-related quality of life and its domains among Australian adults: full sample analysis from the Household, income and labour dynamics in Australia survey (2007–2021) a The regression model includes the grouping variable (wave_count1) that distinguishes consistent and inconsistent participants in model as a random effect to account for the variability between consistent and inconsistent groups, assuming that the individuals in the “consistent and inconsistent” groups are distinct and their HRQoL variation is unique to each person b K10 Kessler Psychological Distress scale (K10). The cut off points of K10 are based on the 2001 Victorian Population Health Survey. The cut-off scores were based on how practitioners use the K10 as a screening tool ( https://www.abs.gov.au/ausstats/[email protected]/ProductsbyReleaseDate/4D5BD324FE8B415FCA2579D500161D57 ) c PD Psychological Distress *, ** show P  < 0.05 and 0.01 respectively HRQoL decreases as the age of the respondents increases (from − 0.016 in 25–44 years to -0.059 in 65 + years). In a similar way, PF, RP, MH, and RE deteriorate with an increase in age. Furthermore, a sharp decline was recorded in PF (0.206) with increasing age. Results show that sex is a strong determinant of HRQoL. Female’s HRQoL decreases more sharply when compared with males. The disutility difference (-0.012) between females and males is substantial. All the four domains also show that females compared to males face higher negative effects. The difference of effect is -0.006, -0.007, -0.012 and − 0.016 in the case of PF, RF, MH, and RE, respectively. The estimated mean HSUs by PD severity and age consistently decline as the age of the respondents increases (see Fig.  1 a and Additional file Table S3a). Sex differences are also visible in mean HSUs with female having sharp decline than males (see Fig.  1 b and Additional file Table S3b). The estimated mean scores of different aspects of HRQoL (PF, RF, MH, RE) by PD severity and age (see Additional file Table S4a, S5a, S6a, S7a and Figure S1a, S2a, S3a, S4a) show similar trends as those of predicted mean HSUs apart from MH (see Additional file Table S6a and Figure S3a), which shows minor improvement after the age of 44 years. The sharp decrease in predicted mean score is reported in case of PF (see Additional file Table S4a and Figure S1a) and RF (see Additional file Table S5a and Figure S2a) especially after the age of 64 years. The predicted mean scores by PD severity and sex demonstrate sex differences across four domains of HRQoL (see Additional file Table S4b, S5b, S6b, S7b and Figure S1b, S2b, S3b, S4b). Overall, the mean predicted scores were higher across the four aspects of HRQoL in case of males than females (see Additional file Figure S1b, S2b, S3b and S4b). Fig. 1 Mean health state utilities among Australian adults by psychological distress severity and age, and by psychological distress severity and sex, based on the household, income and labour dynamics in Australia survey, 2007–2021. a Mean (95% CIs) health state utilites by psychological distress severity and age group. b Mean (95% cis) health state utilities by psychological distress severity and sex Mean health state utilities among Australian adults by psychological distress severity and age, and by psychological distress severity and sex, based on the household, income and labour dynamics in Australia survey, 2007–2021. a Mean (95% CIs) health state utilites by psychological distress severity and age group. b Mean (95% cis) health state utilities by psychological distress severity and sex Marital status emerged as an important determinant of HRQoL and is strongly associated with four domains of HRQoL. Separated/divorced/widowed participants have lower HRQoL when compared with legally married/de facto married participants (-0.018). In contrast, participants who never married/de facto married had the highest HRQoL among all three groups (0.003). Being separated/divorced/widowed also affects PF, RF, MH, and RE. Within these domains, RF (-0.056) and RE (-0.054) were more affected when compared with PF (-0.052) and MH (-0.012). Unlike the above-mentioned finding, no consistent results were found for never married/not de facto married participants in all domains. Level of education was found to be positively associated with HRQoL and its four domains except MH. The higher the education attainment, the higher HRQoL was reported. Participants having an education of 12 years or below had the lowest HRQoL (-0.002). Physical function, RP and RE showed improvement with an increase in level of education. The participants who were unemployed (-0.004) or not in the labour force (-0.018) had a lower HRQoL when compared to participants who were employed. Being unemployed or not in the labour force does not affect PF; however, it negatively affects RF, MH, and RE. Higher levels of physical activity were related to the improved HRQoL of the participants. The higher the level of physical activity, the higher the HRQoL was reported. The utility gain of moderate and high physical activity in terms of HRQoL was 0.013 and 0.021, respectively. Level of physical activity was also positively associated with PF, RF, and RE. In contrast, the association between BMI level and HRQoL and its four domains was negatively reported. Being overweight or obese has disutility of -0.004 and − 0.005 respectively. The highest negative impact of obesity was reported in the domain of RF (-0.039) while obesity was found to have no significant effect on MH. Active club membership proved to be an important determinant of HRQoL. Participants who were active club members have higher HRQoL as compared to non-active members. The estimated health-related disutility attached with non-active members was − 0.008. Active club membership also has a positive effect on PF, RF, MH, and RE. The highest negative impact of non-active club membership was observed in the PF domain (-0.017). Compared to non-smokers, lower HRQoL was associated with smokers. The estimated disutility attached with smokers was − 0.011. Smoking was also found to be adversely affecting PF, RF, MH, and RE. Role emotional was the most adversely affected (-0.022) domain by smoking while the least impact was reported in RF and MH (-0.008). Alcohol consumption did improve the HRQoL of the participants. The utility of drinking increased with the increase in drinking frequency. The estimated utilities associated with occasional, regular, and daily drinkers were 0.006, 0.007, and 0.008, respectively. Drinking was positively linked with PH and RF but had no significant effect on MH and RE. Income gradient proved to be a strong determinant of HRQoL. The rich participants reported a higher HRQoL compared to those who have a low level of income. However, the relationship between HRQoL showed non-linear trend as HRQoL started to decline at the highest level of income (> 200,001). The income was found to be positively associated with all four domains of HRQoL included in the model. However, the highest effect was observed in RF domain (0.058). Australian residing in outer regional and very remote areas have lower HRQoL (-0.002) as compared to those living in major cities. The lowest HRQoL was recorded in RF domain (-0.021) of HRQoL. The two background variables included in the model, Indigenous status and English speaking did not show any significant association with HRQoL or any of its domain. In addition to our main model, we performed two more sets of regression by dividing our samples into two groups, i.e., participants that information was available across all 8 waves and those whose information was available on less than 8 waves. We found consistency of findings across both sub-samples with minor exceptions (see Additional file Table S8a and Table S8b).

Data on SF-36 responses, and other sociodemographic characteristics, including age, sex, English speaking, indigenous status, region of residence, marital status, education, employment, physical activity, body mass index (BMI) level, active club membership, smoking, drinking and income of individuals detected with PD aged ≥ 15 years were sourced from the 8 waves (7, 9, 11, 13, 15, 17, 19, 21) of the nationally representative HILDA survey spanning the years 2007 to 2021. The selection of waves is based on the availability of data on our key exposure variable, the PD. The Kessler Psychological Distress Scale (K10) commonly used to detect PD, was first included in the survey in wave 7 and has subsequently been included biennially [ 18 ]. The survey examines the same participants repeatedly over time making it advantageous to other repeated cross-sectional survey data. Furthermore, assessing same participant repeatedly overtime enables each participant to control trends in prevalence of PD [ 39 ]. Approximately, the percentage of reinterviewed participants at each wave is 90%, and of which 85–90% return the self-completed questionnaire (SCQ) which is completed individually by respondents, while other questionnaires, such as the Household Form (HF) and Household Questionnaire (HQ), are typically administered by interviewers. Other surveys of a similar nature have similar attrition [ 40 ]. Due to long-term engagement of households in the study, which may not participate in traditional single person studies, and their willingness to report mental health condition because of self-complete mode of K10, the greater accuracy in participants responses is expected [ 41 ]. The longitudinal design of the HILDA survey also allows the possibility of returning of individuals in subsequent waves who declined to participate at one point [ 39 ]. Our study sample includes individuals reporting PD from the general population rather than recruited from outpatient clinics or disease-specific cohorts. The analysis included five outcome variables: HSUs, PF, RP, MH, and RE. In health care, HRQoL is an increasingly important and useful indicator of overall health [ 42 , 43 ] with an exclusive focus on one’s quality of life considering optimal functioning of physical and mental health [ 44 , 45 ]. HSUs quantify health states severity derived from people’s own preferences and are the quality component of quality adjusted life years (QALYs) which combine length and quality of life into a single metric [ 46 ]. In rationing decisions and cost-effective studies, use of QALYs is common [ 47 ]. Based on rating scale, there are multiple measures to assess HRQoL. Short form-36 has been widely used to capture physical, social and mental HRQoL due to its validity and reliability in general and depressed populations [ 48 – 51 ]. We assessed HRQoL using HSUs derived from SF-36 utilising SF-6D algorithm. Psychological distress and its various classifications were the main exposure variable in this study. We used K10 scale to detect PD, as it is increasingly applied in primary care and mental health settings as both a screening tool and an outcome measure [ 52 – 55 ]. The K10 is preferred over other measures such as Short Form Survey-12 because it was designed as a measure of distress not disability [ 56 ]. There is no agreement regarding the cut-off scores of the K10 to screen PD. We follow [ 20 ] and use cut-off scores for the K10 as 10–19 for “no PD”, 20–24 “mild PD”, 25–29 “moderate PD” and 30–50 “severe PD”. Based on our understanding of HRQoL literature [ 26 , 28 , 29 , 31 , 36 , 57 – 63 ], other sociodemographic and socioeconomic determinants included in the analysis are age, sex, English speaking, indigenous status, region of residence, marital status, education, employment, physical activity, BMI level, active club membership, smoking, drinking and income. Table 1 provides descriptions of all variables used in the study. Table 1 The definition of the variables used in the analysis Variable Abbreviation Definition SF-6D Health state classification Health state utilities Continuous variable ranges from 0 to 1 SF-36 Physical functioning Physical function Continuous variable ranges from 0 to 1 SF-36 Role-physical Role physical Continuous variable ranges from 0 to 1 SF-36 Mental health Mental health Continuous variable ranges from 0 to 1 SF-36 Role-emotional Role emotional Continuous variable ranges from 0 to 1 Kessler Psychological Distress Scale (K10) score Psychological distress Categorical variable scored 1 for no psychological distress (K10: 10–19), 2 for mild psychological distress (K10: 20–29), 3 for moderate psychological distress (K10: 25–29), 4 for severe psychological distress (K10: 30–50), 5 for Missing/Undetermined Age last birthday on June 30 2007 Age Categorical variable scored 0 for 15–24 years, 1 for 25–44 years, 2 for 45–64 years, 3 for 65 + years Sex Sex Binary variable scored 0 for males and 1 for females Aboriginal or Torres Strait Islander origin Indigenous status Categorical variable scored 0 for Not of indigenous origin, 1 for Aboriginal or/and Torres Strait Islander,2 for Missing/Undetermined Speak language other than English Speak language other than English Categorical variable scored 0 for Yes, 1 for No 2 for Missing/Undetermined Remoteness area Remoteness area Categorical variable scored 0 for Major Cities, 1 for Inner Regional, 2 for Outer Regional or Remote or Very. Remote 3 for Missing/Undetermined Marital status Marital status Categorical variable scored 0 for Legally Married & De Fecto, 1 for Separated, Divorced and Widowed, 2 for Never married and not de facto, 3 for Missing/Undetermined Highest education level achieved Education level Categorical variable scored 0for Year 12 and below, 1 for Cert III or IV, 2 for Bachelor/honours and diploma/adv diploma, 3 for Postgrad, grad diploma & grad certificate, 4 for Missing/Undetermined Current labour force status Current labour force status Categorical variable scored 0 for Employed, 1 Unemployed, 2 for Not in labour force, 3 for Missing/Undetermined Categorical Physical Activity Physical activity level Categorical variable scored 0 for Low, 1 for Moderate, 2 for High, 3 for Missing/Undetermined Body Mass Index (BMI) Body Mass Index Categorical variable scored 0 for Normal, 1 for Overweight, 2 for Obesity 3 for Missing/Undetermined Currently an active member of a sporting/hobby/community-based club or as Club membership Categorical variable scored 0 for Yes, 1 for No 2 for Missing/Undetermined Smokes cigarettes or other tobacco products Smoking status Categorical variable scored 0 for No, 1 Yes, 2 for Missing/Undetermined Drink alcohol Alcohol Consumption Categorical variable scored 0 for Abstinent, 1 for Daily or Almost Daily: >5 days per week, 2 for Regular: 2–4 days per week, 3for Occasional: <2 days per week, 4 for Missing/Undetermined Financial year disposable total income ($) Annual disposable income Categorical variable scored 0 for $0-$30,000, 1 for $30,001-$60,000, 2 for $60,001-$100,000, 3 for $100,001-$200,000, 4 for >$20,0001 The definition of the variables used in the analysis

The life course theory provides an integrative framework for understanding how both micro and macro level factors affect HRQoL [ 38 ]. The theory states that an individual life at different stages is intricately intertwined with past and present factors. It considers the dynamic interplay of biological, psychological, and social factors in shaping health outcomes over time. When applied to the concept of HRQoL, life course theory provides insights into how different life experiences and transitions can impact an individual’s overall well-being. The differential impact of PD on HRQoL may be elucidated through theoretical pathways, positing that variations in HRQoL, as influenced by PD, are predominantly driven by three main factors as follows: 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:HRQoL=F\left(PD,\:D,\:SE\right)$$\end{document} Where HRQoL (or its domains, PH, RP, MH, RE) is modeled as a function of PD, sociodemographic factors (D) such as age, sex, area of residence, and socioeconomic factors (SE) such as education, employment status and income level. In this study, our three main objectives are: (i) to estimate causal effects of different classifications of PD on HRQoL, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{dHRQoL}{dPD}$$\end{document} , (ii) to document to what extent different sociodemographic factors affect HRQoL ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\frac{dHRQoL}{dD}\&\;\frac{dHRQoL}{dSE}$$\end{document} ), and (iii) to examine the role of time as a sole determinant of HRQoL of people with PD.

This study contributes to our understanding of psychological distress-HRQoL gradient. We draw upon data from nationally representative survey of Australian adults to investigate the association between PD and HRQoL in longitudinal settings. We also identify sociodemographic and socioeconomic factors that are closely related with HRQoL. Additionally, we examine whether the effect of PD and its different classifications impact different aspects of PD proportionately. We present new evidence that after controlling for other sources of variation, time itself plays a less direct role in HRQoL determination while age, sex and PD severity play more influential roles in determining HRQoL. Furthermore, PD and its different categories are negatively associated with HRQoL. We also observed that PD affects different aspects of HRQoL disproportionately with a lower quality of life in the domains of MH and RE. Our finding shows that because of PD, HRQoL of females decreases more than males while HRQoL declines with an increase in age. Level of education, being married, active club membership, physical activity, being employed, drinking, level of income were positively associated with PD while BMI level, and smoking were found to be negatively affecting HRQoL. Indigenous status, English speaking and remoteness of area were not statistically associated with HRQoL. From a policy perspective, it is important to understand the considerable variance and nuance in the effects of PD on HRQoL and its domains. To improve HRQoL of people with PD, health interventions should be tailored to the severity levels, and other factors e.g., age and sex of patients, with more attention given to females and older individuals. Moreover, a coordinated effort across different sectors, involving government agencies, healthcare providers, educators, employers, and communities is required. Additionally, investing in community-based PD programs to provide support and resources at the local level can also be of help. Such coordinated and customized approaches can optimize the effectiveness of treatment and prevention strategies of PD. This study further emphasizes policy attention to increasing the quality of life of people with PD by illuminating the key role of the demographic and socioeconomic factors in determining HRQoL and its domains.

The primary aim of this study was to provide longitudinal evidence on the association between different classifications of PD and HRQoL including its domains of PF, RP, MH, and RE in Australian adults. Additionally, socioeconomic and sociodemographic determinants of HRQoL and its various domains were investigated. We found that different classifications of PD affect HRQoL and its various domains disproportionally. Education, physical activity, employment, alcohol consumption, and income appeared to act as protective factors against PD, whereas elevated BMI level and smoking were detrimental to HRQoL. The findings from this study provide valuable insights that can inform clinical practice and health policy, supporting the development of more targeted interventions and tailored approaches for the effective management of PD in Australia or similar population. We found that PD and its different classifications adversely affect HRQoL. This finding is in line with previous studies that report negative association between PD and HRQoL [ 60 , 92 ]. PD impairs proper functioning of life tasks including bathing, dressing, driving, and budgeting, which compromises HRQoL [ 93 ]. Severe PD significantly interferes with routine activities by causing functional disability that has negative influence on HRQoL [ 94 – 96 ]. In Spanish and French studies, PD has been linked to low HRQoL [ 97 ]. People with PD are more likely to report poor HRQoL due to decreased energy and motivation level, depressed mood, and pessimistic outlook [ 98 ]. Unfavorable perceptions of their circumstances on the part of individuals with PD could limit their ability to fairly perceive HRQoL. This reinforces the need and warrants further attention to address PD in the Australian and related populations and to acknowledge potential ramifications that PD may have on the HRQoL in the long run. This will help in choosing more effective intervention and determining more appropriate strategies regarding public resource allocation. On closer examination, the observed changes in HRQoL over time appear to be attributable not only to the passage of time but also to the influence of various SES factors. Our simple models (see Additional file Table S1&S2) show a negative association between time and HRQoL with relatively larger and statistically significant coefficients for time. However, as we incorporate other determinants in the model (Table 3 ), both fixed and random effects, a remarkable shift occurs and the coefficients of time reduce in magnitude, almost approaching to zero, while age, sex, PD classification and some other variables emerge as prominent and statistically significant determinants of HRQoL. This finding contrasts with those studies that have reported decline in HRQoL over time [ 99 , 100 ]. A study conducted in UK showed a decline in HRQoL in a general population over a 5-year observation period and the result were unchanged even after adjusting for potential confounders [ 101 ]. Concerning our findings of the insignificant role of time in determining HRQoL, no former study reported a corresponding result to our knowledge so far. However, we assume that this finding may be due to the following two reasons. First, due to improvement in medical care in Australia, life expectancy has increased which is not accompanied by morbidity as the case in other countries such as UK. Secondly, expectations and perceptions regarding HRQoL remain unchanged, and are not associated with the perception towards medicalization of conditions. This underscores the need for policies that prioritise socioeconomic and behavioural determinants over temporal factors, directing resources toward modifiable risks to improve HRQoL in individuals with PD. Given Australia’s vast geography, remoteness remains a significant concern. This study results show that remoteness was negatively associated with HRQoL and its various domains excluding role emotional which is in line with previous studies conducted in Australia. One study found a small stepwise gradient in utility by remoteness [ 102 ]. City dwellers expected an additional 3.8 quality adjusted life years to those living in outer regional or remote areas [ 103 ]. This could be attributed to the fact that participants in remote regions were older compared to those in major cities, a pattern also reflected in other datasets such as those reported by the Australian Institute of Health and Welfare [ 104 ]. Our finding expands our understanding of HRQoL disparities associated with regional status in the Australian population and helps inform the development of targeted health policies to better address the issue of PD, with particular attention to remoteness, as lower HRQoL was observed in remote areas of Australia. When examining sex differences, HRQoL was found to be significantly lower in all age-groups of females. This finding is consistent with previous studies stating that women report lower HRQoL than men [ 105 ]. A multitude of possible explanations can elucidate this gender inequality. Female diseases such as endometriosis which are common especially in young females leads to decrease in HRQoL [ 106 ]. Some sex differences may be explained by female’s menopause which affects HRQoL mainly through PF domain of SF-36 [ 107 ]. Caregivers of older relatives often experience reduced HRQoL, and since informal caregiving is predominantly undertaken by females, this may help explain the lower HRQoL observed among females in our sample compared with males [ 108 ]. Furthermore, it is also well-documented in the literature that despite having higher level of education, females often have poorer economic conditions than males [ 29 ]. This indicates the presence of gender-specific trajectories, which future research could investigate using mixture modelling to determine whether these patterns follow a normal distribution or whether distinct subgroups emerge across gender. We found that aging has adverse effects on HRQoL especially for those enduring PD. Prior studies have suggested that increasing age negatively affects HRQoL [ 109 , 110 ]. A study revealed that in Europe young age was associated with better HRQoL [ 111 ]. Another study found that older adults are more likely to report lower HRQoL when compared with young ones [ 112 ]. Risk of morbidity and mortality increases due to PD in later age and this increases likelihood of inclinations to commit suicide [ 113 ]. Significant physical and biological changes occur in old age that inevitably have implications in terms of HRQoL. Due to many physical and mental disturbances including chronic disease pain, decreased cardiac output, progressive elevation of blood glucose, loss of muscle mass, functional disability and loneliness can decrease HRQoL of elderly [ 114 – 116 ] as body’s capacity to adapt to internal and external stressors diminishes, which can result in decompensation and the clinical manifestations of frailty, including impaired mobility, increased vulnerability to confusion, and a heightened risk of functional deterioration [ 117 ]. Although age is an important predictor of HRQoL [e.g [ 118 ]. , , its relationship with HRQoL is complex, as studies have shown that quality of life and its specific domains vary considerably across individuals and groups [e.g [ 112 ]. , Therefore, it is essential to consider a range of other potential confounding factors when examining this association [e.g [ 119 ]. , . Our findings also show that aging has stronger impact on PF and RP as compared to MH and RE. Many factors explain this finding. First, physical health requires functional mobility and strength. With aging, due to many biological factors, body strength decreases which hampers physical functionality to be performed in a satisfactory way and hence reduces HRQoL. Studies found that mental health further accentuates this association by increasing the complexity of the problem [ 120 ]. Second, older adults with PD suffer isolation and loneliness which are an outcome of limited functional mobility [ 121 ]. With increasing age, social support, a buffer against PD, declines which tends to make people fight against emotional conditions by providing a network for practical assistance and help when needed [ 122 ]. A previous Norwegian study showed that among the elderly, PD negatively affects functional ability [ 123 ] and hence HRQoL. The observed patterns show that physical dimensions of HRQoL are more vulnerable to the effects of aging than mental and emotional aspects. This suggests that greater emphasis should be placed on maintaining physical functioning and mitigating role limitations while continuing to support psychological resilience in later life. We found a clear relationship between high HRQoL and higher level of education which is indisputable at both the individual and public investment level [ 124 ]. Higher level of education makes possible the adoption of healthier lifestyles through better access and use of health information and resources [ 125 ]. Health complications in early age hinder academic performance that subsequently reduce employment opportunities leading to future resource constraints making access to better health resources relatively difficult [ 126 , 127 ]. From a materialistic perspective, the adverse effects of deprivation and social exclusion indicate that lower educational attainment is associated with poorer health outcomes. This perspective emphasizes the social consequences of poverty, such as disparities in life expectancy [ 128 , 129 ]. In formal education, credentialism–the attainment of degrees and certifications recognized by the education system–often translates into improved employment prospects, higher income, and greater social standing, all of which can contribute to an enhanced overall quality of life and long-term socioeconomic well-being [ 130 ]. In prior studies, more years of education were significantly found to be negatively associated with worse future HRQoL [ 109 ], which is consistent with our findings. These findings underline the importance of education in relation to HRQoL and highlights how this construct associates to socioeconomic characteristics of the population [ 131 ]. Implementing targeted public policies aimed at enhancing educational opportunities can yield benefits beyond academic achievement and may also contribute to better HRQoL, highlighting the broader societal impact of investing in education. We found that being legally married/de facto married increases HRQoL as compared to separated, divorced and widows. The never-married category was found to have the highest HRQoL. In earlier studies, similar findings were observed [ 132 ]. The increased social burden such as committing family responsibilities associated with marriage could explain why being single has greater HRQoL in our study. This burden restricts proper care of children and hinders proper functioning of duty; hence, married individuals prospect their HRQoL negatively. Additionally, married people also face problems like reinterpretation of mutual life plans, challenging marital intimacy, adaptation to familial and partnership responsibilities [ 133 ]. Other studies found different results and associated being married with better HRQoL [ 134 ] as they feel cared for, and affectionate through good relationship with their spouses and families which makes an individual feel esteem and worthwhile [ 36 ]. Furthermore, expression of emotions with partner and receiving enough social support creates a promising outlook on life. The variation in findings requires further research on the social and cultural dimensions of marital status. Our findings show that employed individuals score higher on HRQoL than the unemployed and those who were not in the labour force. Several prior studies documented the positive association between employment and HRQoL [ 60 , 135 ]. Employed people have large social network and enjoy higher self-esteem which enhances HRQoL [ 98 ]. Additionally, employed individuals have higher resources at their disposal to cover their needs than people without job or out of the labour force [ 14 ]. This finding highlights the protective role of employment in sustaining HRQoL underscoring the need for supportive policies that encourage workforce participation among individuals at risk of poorer health outcomes. Our findings of the positive association between physical activity and a higher level of HRQoL were in line with previous studies conducted in Australia [ 136 – 138 ]. Similar association between HRQoL and physical activity were reported in studies conducted in China [ 61 ], Spain [ 139 ], the UK and elsewhere [ 140 , 141 ]. A likely explanation is that regular physical activity improves physical fitness and functioning while fostering a greater sense of control [ 142 – 144 ]. Additionally, physical activity improves HRQoL by reducing chronic inflammatory levels [ 145 ], enhancing considerable mental simulation [ 143 ], and decreasing stress and negative emotions [ 146 ]. This finding provides strong evidence on the role of physical activity in improving HRQoL underscoring the need for its integration into health promotion strategies. Results from our study have also indicate that obesity adversely affects HRQoL. This finding is in agreement with the prior studies which showed a decrease in HRQoL because of the increase in BMI level [ 15 , 147 , 148 ]. This finding is explained in the following way. Unhealthy dietary behaviour e.g., high intakes of sugar, fats, and fast-food causes obesity and negatively influences HRQoL [ 149 , 150 ]. Nonetheless, some studies mention unclear associations between BMI level and HRQoL due to the complex nature of the relationship. For some individuals, BMI level could be primary indicator of HRQoL, while for others the association may not exit at all or be minimal [ 26 ]. In general, studies assuming linear association between the two variables find the adverse effect of BMI level on HRQoL [ 151 , 152 ] while studies assuming non-linear relationship do not necessarily endorse this association. Furthermore, the impact of obesity on HRQoL is disproportionate for population subgroups. For example, obesity effects on HRQoL varied by age and sex. Obesity found to greatly affect women than their male counterparts. Differential impacts of BMI on HRQoL were recorded for different age groups [ 153 – 157 ]. Because of the complex nature of the association between BMI level and HRQoL, our study points to the need for future research aimed at further clarifying this relationship. The observed association between active club membership and a more favourable HRQoL is in line with those of previous studies [ 158 – 162 ]. Engaging in sports activities fosters physical condition [ 163 , 164 ], and hence HRQoL. Organized sports provide valuable opportunities for social interaction and peer engagement which foster positive social experiences and enhance social skills, contributing to improved HRQoL [ 165 – 167 ]. Membership of various community groups such as clubs and societies promotes well-being in older age [ 168 ]. Significant changes in various dimensions of well-being were observed in previous interventional studies involving engagement in community arts and music groups [ 169 , 170 ]. This suggests that participation in community activities can positively influence well-being, emphasizing the need for policies that strengthen family, workplace, educational, and religious communities, as these domains are closely tied to human flourishing [ 171 ]. As expected, in this study adverse effects of smoking on HRQoL were observed. This finding is confirmed by prior cross-sectional [ 172 , 173 ] and longitudinal studies [ 174 , 175 ] focusing on the association between smoking and HRQoL [ 110 ]. Several possible interpretations of this result can be stated. First, smoking can lead to increased odds of depression and clinically significant fatigue [ 176 , 177 ]. Second, smoking is associated with risk of many diseases including cardiovascular and cancer [ 178 ], which results in worse HRQoL [ 179 ]. Third, substances inhaled in tobacco are notorious for creating weakness of muscle and decreased vitality [ 180 ]. This finding provides compelling evidence to quit smoking [ 181 ] and contrasts to the popular belief that smoking related pleasure will lost due to quitting. However, there are studies that did not observe any association between smoking and HRQoL [ 15 , 182 ]. Given the suggested benefits of not smoking for HRQoL verified in this study, promoting smoking cessation should remain a central public health priority. Programs aiming strengthening prevention and cessation could lead to substantial improvements in population well-being and reduce the burden on health systems. Our findings also show that participants who drink reported higher HRQoL as compared to those who did not drink which contradicts the belief where drinking is considered a poor lifestyle behavior that negatively affects HRQoL and is some time adopted to alleviate stress [ 183 ]. Several cross-sectional studies reported results consistent to our findings. Moderate alcohol consumption increases HRQoL [ 184 ] while excessive use of alcohol and binge drinking results in poor HRQoL [ 185 ]. Longitudinal studies show that moderate drinking improves mental and physical health [ 186 ]. Compared to abstaining or decreasing use of alcohol, persistent and moderate drinking has been observed to be pro mental and physical HRQoL [ 187 ]. However, there are studies that report findings inconsistent with our findings. In a Turkish study, drinking has been shown to negatively affect HRQoL [ 188 ]. In Serbia, individuals who abstained from drinking exhibited a higher HRQoL compared to those who consumed alcohol [ 189 ]. While we record a potential benefit of drinking on HRQoL, the evidence remains inconsistent, with some studies reporting adverse effects of drinking. Therefore, we suggest caution in interpreting this finding, as the impact of alcohol consumption may depend on other sociocultural and behaviour and individual health factors. For balanced public health recommendations, future research should disentangle these complexities. The study results demonstrates that HRQoL improves as the level of income increases, which is consistent with the prior findings [ 63 , 190 ]. This finding is understandable as individuals with high income have better access to health services and material goods and have high HRQoL as a result [ 191 ]. In contrast, people with lower income are more likely to be exposed to risks such as stress, drinking, smoking and consequently have lower levels of HRQoL. Therefore, when designing interventions such as greater investment in health and related services to reduce health inequalities, it is essential to account for the income gradient to ensure equitable improvements in HRQoL. In general, indigenous status and English speaking were not found to be statistically significant determinants of HRQoL and its aspects however, they were used in the analysis as a background variable and their relevance to impact HRQoL independently [ 94 ]. Additionally, the associations between the predictors examined in this study and overall HRQoL mirrored their effects across specific HRQoL domains, reinforcing the consistency and robustness of these relationships. This study has several notable strengths. First, we draw upon a large sample that is nationally representative for the Australian with a response rate par with many international surveys. Second, this study identifies several demographic and socioeconomic factors that can determine changes in HRQoL. Third, we examined the effect of PD and its different classifications on various domains of HRQoL to identify if this effect is proportionate across different domains. Finally, and more importantly, the study investigated the role of time in the context of HRQoL of people with PD as a sole determinant. Our findings are not without limitations that must be taken into account. First, many of the variables used in the analysis are subjective due to self-reporting. This could create a response bias. Nonetheless, in a wide range of populations, self-reported measures are commonly used due to their validity and reliability [ 192 ]. For example, due to its brevity and validity K10 scale is considered to be an appropriate proxy for the actual PD used in the literature. Second, despite this study accounting for determinants of HRQoL, the list is not exhaustive. Based on data availability, the inclusion of several other determinants e.g., different types of dietary behaviour, electronic cigarette use in the analysis can be useful. Future research should address these limitations. Moreover, new studies need to focus on the trajectories and transmission probabilities and the economic cost of PD in terms of HRQoL. Clinically diagnosed measures of PD may be used to substantiate the relationship between PD and HRQoL. Despite these limitations, this study unfolds longitudinal information regarding a central aspect of PD and HRQoL relationship that is key to policy formulation.

To establish causal relationship between PD and HRQoL, we regress HRQoL and its different domains on PD and a set of socioeconomic, sociodemographic, and clinical factors. The HRQoL function described in Eq.  1 can be written as five separate econometric models as: 2 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{HRQoL}_{it}={\beta\:}_{1}+{\beta\:}_{2}{T}_{it}+{\beta\:}_{3}{PD}_{it}+{\beta\:}_{4}{D}_{i}+{\beta\:}_{5}{SE}_{it}+\epsilon_{it}.$$\end{document} 3 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{PF}_{it}={\beta\:}_{1}+{\beta\:}_{2}{T}_{it}+{\beta\:}_{3}{PD}_{it}+{\beta\:}_{4}{D}_{i}+{\beta\:}_{5}{SE}_{it}+\epsilon_{it}$$\end{document} 4 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{RF}_{it}={\beta\:}_{1}+{\beta\:}_{2}{T}_{it}+{\beta\:}_{3}{PD}_{it}+{\beta\:}_{4}{D}_{i}+{\beta\:}_{5}{SE}_{it}+\epsilon_{it}.$$\end{document} 5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{MH}_{it}={\beta\:}_{1}+{\beta\:}_{2}{T}_{it}+{\beta\:}_{3}{PD}_{it}+{\beta\:}_{4}{D}_{i}+{\beta\:}_{5}{SE}_{it}+\epsilon_{it}.$$\end{document} 6 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{RE}_{it}={\beta\:}_{1}+{\beta\:}_{2}{T}_{it}+{\beta\:}_{3}{PD}_{it}+{\beta\:}_{4}{D}_{i}+{\beta\:}_{5}{SE}_{it}+\epsilon_{it}.$$\end{document} Where HRQoL, the primary outcome variable, is the HRQoL of individual \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:i\:$$\end{document} at time \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:t$$\end{document} . PF, RF, MH, and RE represent physical function, role physical, mental health, and role emotional of individual \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:i$$\end{document} at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:t$$\end{document} , respectively \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:.{D}_{i}$$\end{document} is a vector of time invariant demographic variables and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{SE}_{it}$$\end{document} is vector of socioeconomic variables that can affect HRQoL. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{T}_{it}\:$$\end{document} measures time for individual \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:i$$\end{document} at \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:t$$\end{document} . \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\epsilon_{it}$$\end{document} represents unobservable determinants of HRQoL. The parameters of the model (the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\beta\:}^{{\prime\:}S})$$\end{document} are commonly referred as fixed effects. The error term in Eq. ( 2 ) has two components. 7 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\epsilon_{it}=\gamma\:+{\delta\:}_{it}.$$\end{document} Where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\gamma\:\sim\:i.i.d.(0,\:{\sigma\:}_{u}^{2})$$\end{document} is an individual specific component that captures time-invariant unobserved components factors. The \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\delta\:}_{it}\sim\:\text{N}(0,{\sigma\:}_{\delta\:}^{2})$$\end{document} is an individual specific time varying component of the error term which captures the impact of other unobserved variables that can potentially affect HRQoL of an individual. To get consistent estimates of the parameters (the β^’s) in Eq. ( 2 ), it is assumed that there are no unobservable factors that systematically coincide with independent variables and influence the outcome variable. To strengthen plausibility of this assumption [ 64 ] and to increase precision of estimates [ 65 ], we follow the standard practice in the following way. First, we adjust our analysis for imbalances in some baseline characteristics even though the model specification in Eq. ( 2 ) can potentially adjust for baseline differences. Second, we add some common covariates in the model as determinant of HRQoL. We estimated all models by using LMM with a longitudinal design. The key strength of using LMMs lies in their ability to identify whether the gap in HRQoL among individuals with different PD severity levels, sexes, or age groups is narrowing or widening over time or it is staying the same. The LMM can overcome the effects of missing data countered in the analysis as we did not restrict our analysis to respondents with complete data, a popular approach to handle missingness which causes some loss of information, to avoid having biased and inefficient estimates [ 66 – 68 ]. Based on the multivariate normal distribution, LMM incorporates all available observations through likelihood-based estimation to make efficient use of partially observed data instead discarding cases with missing values to provide valid estimates. This approach enhances both statistical precision and power. Additionally, the LMM preserves coverage probabilities and control the false positive rate more effectively than traditional methods such as complete case analysis and mean imputation, as they explicitly account for the correlation structure within the data and the associations among variables [ 69 , 70 ]. Moreover, in longitudinal analysis, incorrect specification of the covariance structure for repeated measures may result in biased regression estimates and either underestimation or overestimation of the associated standard errors when data are missing. The LMM incorporates an unstructured covariance matrix for the repeated measures [ 71 ] which is recommended by several authors since it is straightforward to pre-specify, involves minimal power loss compared with more parsimonious options [ 71 – 73 ], and yields estimates that are approximately equivalent to, and in some cases slightly more efficient than, those obtained through comparable multiple imputation methods procedures [ 72 , 74 ]. Furthermore, LMM are increasingly used to obtain longitudinal outcomes [ 75 ] as they can be tailored according to data features and their results are easily reproduceable. Another feature of the LMM is its ability to produce robust estimates from non-normally distributed data. As compared to traditional repeated-measure data analysis, LMM provides greater flexibility by accommodating both time-dependent and time-independent covariates within the model. This method has been found to be working well in small sample sizes [ 76 ]. The prior studies mainly used ordinary regressions or multivariable Analysis of Variance (ANOVA) in studying the effects of different determinants of HRQoL without accounting for the effects of repeated measures and missing data [ 77 ]; arguably our analytical framework more effectively addresses the related problem. In LMM parameter estimation is obtained by maximizing the likelihood of the full data [ 78 ], which is analogous to classical linear models. To obtain final estimates, this method relies on iterative procedures such as Newton–Raphson [ 79 ], Average Information [ 80 ], Restricted Maximum Likelihood [ 78 ] and Expectation–Maximization [ 81 ] algorithms. Overall, LMM has proven to be one of the most effective methods for managing incomplete data [ 72 ]. Our decision to model the SF-36 domains as separate dependent variables to assess the impact of PD was guided by prior research [ 82 – 86 ] and by the theoretical framework of the SF-36, in which each domain is treated as a distinct dimension of health status. This approach also helps to mitigate the risk of multicollinearity that could arise from the substantial correlations often observed among the SF-36 domains if they were simultaneously included as independent variables or alongside the overall HRQoL score. Multicollinearity can inflate and destabilize standard errors, which in turn produces highly unreliable p-values when testing the significance of predictors. This instability can ultimately lead to misleading inferences and implausible interpretations of the model results [ 87 – 90 ]. Additionally, in the presence of multicollinearity interpretation of regression coefficients as the effect of a small change in that variable while holding other constants becomes practically impossible [ 91 ].

Mental health conditions affect one billion people worldwide, with profound consequences on life expectancy, economic burden, and overall well-being [ 1 ]. Life expectancy can be reduced by up to 20 years due to premature physical illness and suicide associated with mental health conditions. The global annual cost associated with mental illness is projected to reach $6 trillion by 2030 [ 2 ]. Two common mental disorders, depression and anxiety, alone can cause chronic physical diseases including dementia [ 3 ], cancer [ 4 ], and cardiovascular disease [ 5 ], and all can lead to mortality [ 6 ] along with causing years lived with disability. The mental health decline among Australian adults, particularly in youth (16–24 years), is alarming, with a 50% increase in mental disorders from 2007 to 2019 [ 7 – 9 ]. Currently, mental illness is the leading cause of disability and chronic disease in Australia [ 8 , 10 ]. The rise in young females was even greater than in young males (48%). The Household Income and Labour Dynamics in Australia (HILDA), a nationally representative survey showed a consistent decline in mental health of this age group. The increasing prevalence of mental health issues imposes significant human and economic burden. In 2020-21, Australia spent $A11.6 billion (A$451 per person) on mental health-related services, highlighting the substantial economic burden [ 11 ]. Psychological distress (PD), a form of mental disorder, prevalent among both adults and seniors, poses a significant public health challenge and is a major contributor to the global burden of disease [ 12 , 13 ]. Growing evidence indicates that PD is associated with various adverse outcomes including diminished health-related quality of life (HRQoL) [ 14 – 18 ]. Through a combination of physical illness and suicide, severe PD can reduce life expectancy by up to 20 years [ 2 ]. Women with PD bear more health care cost than women without PD [ 19 ] and HRQoL decreased more steeply with an increase in age among individuals experiencing PD compared to those without PD [ 20 ]. In Australia, the PD burden of employee productivity and time loss from work was estimated to be $5.9 billion and 50–60%, respectively [ 21 , 22 ]. PD has various severity levels that can impact HRQoL and its domains in different proportions such as physical function (PF), role physical (RP), mental health (MH) and role emotional (RE) [ 23 , 24 ]. Understanding how the PD severity classifications impact HRQoL and its domains is important for effective and efficient allocation of scarce health care resources across competing disease groups as they offer insights into the intensity of individuals’ preferences for various health states [ 25 ]. Furthermore, to help clinicians and policy makers to establish a long-term management plan for PD, it is important to investigate socioeconomic and demographic factors that influence the HRQoL of individuals experiencing PD. Previous studies on HRQoL often focused on certain disease, situation or investigated the cost-effectiveness of interventions and care programs [ 16 , 26 – 33 ]. However, the impact of different classifications of PD on HRQoL in representative sample of adults, where HRQoL may have varying patterns, remains largely unexplored. Despite several studies on the association between PD and HRQoL [ 34 – 37 ], there is a dearth of studies that use longitudinal data and linear mixed model (LMM) in such examinations, hence unable to assess whether the disparity in HRQoL among individuals with varying PD severity levels is either contracting, expanding, or remaining constant over time. It is also unclear whether the different levels of PD severity impact various aspects of HRQoL to the same extent. Furthermore, we investigate the association between several sociodemographic factors associated with HRQoL of individuals with PD. To our knowledge, no previous research has addressed the specific aspects of PD and HRQoL investigated here. The main aim is to investigate the nuanced interplay between different PD classifications and HRQoL, along with its various domains, among Australian adults using 16 years of longitudinal data. This study makes significant contributions to existing literature on the association between HRQoL and PD in several ways. First, it provides crucial longitudinal evidence, distinguishing itself from previous cross-sectional studies by identifying determinants of HRQoL, including PD classifications, age, and sex, using nationally representative data from the HILDA survey. Second, the paper examines the domain-specific impact of different PD classifications on HRQoL, analyzing the sensitivity of PF, RF, MH, and RE. Third, the innovative LMM incorporates both fixed and random effects, providing a comprehensive understanding of population-level trends and individual-specific variations. Lastly, the study’s insights on the association of time with HRQoL have potential applications in PD-related cost-effective analyses, allowing for the incorporation of time-sensitive HRQoL weights in evaluating the effectiveness of health interventions.

Additional file 1: Table S1: Linear mixed effects models with time as a predictor of health-related quality of life and its domains among Australian adults: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S2: Linear mixed effects models with time and psychological distress severity as predictors of health-related quality of life among Australian adults: full sample analysis from the Household, Income and Labour dynamics in Australia survey (2007–2021). Table S3a: Predicted mean health state utilities of study respondents with psychological distress by age group: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S3b: Predicted mean health state utilities of study respondents with psychological distress by sex: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S4a: Predicted mean physical function scores of study respondents with psychological distress by age group: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S4b: Predicted mean physical function scores of study respondents with psychological distress by sex: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S5a: Predicted mean role physical scores of study respondents with psychological distress by age group: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S5b: Predicted mean role physical scores of study respondents with psychological distress by sex: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S6a: Predicted mean mental health scores of study respondents with psychological distress by age group: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S6b: Predicted mean mental health scores of study respondents with psychological distress by sex: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S7a: Predicted mean role emotional health scores of study respondents with psychological distress by age group: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S7b: Predicted mean role emotional health scores of study respondents with psychological distress by sex: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Fig. S1a & S1b: Mean physical function scores among Australian adults by psychological distress severity and age, and by psychological distress severity and sex, based on the Household, Income and Labour Dynamics in Australia Survey, 2007–2021. Fig. S2a & S2b: Mean role physical scores among Australian adults by psychological distress severity and age, and by psychological distress severity and sex, based on the Household, Income and Labour Dynamics in Australia Survey, 2007–2021. Fig. S3a & S3b: Mean mental health scores among Australian adults by psychological distress severity and age, and by psychological distress severity and sex, based on the Household, Income and Labour Dynamics in Australia Survey, 2007–2021. Fig. S4a & S4b: Mean role emotional scores among Australian adults by psychological distress severity and age, and by psychological distress severity and sex, based on the Household, Income and Labour Dynamics in Australia Survey, 2007–2021. Table S8a: Linear mixed effects models with predictors of health-related quality of life and its domains among respondents with complete data for all 8 waves: analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S8b: Linear mixed effects models with predictors of health-related quality of life among respondents with data for fewer than 8 waves: analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Additional file 1: Table S1: Linear mixed effects models with time as a predictor of health-related quality of life and its domains among Australian adults: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S2: Linear mixed effects models with time and psychological distress severity as predictors of health-related quality of life among Australian adults: full sample analysis from the Household, Income and Labour dynamics in Australia survey (2007–2021). Table S3a: Predicted mean health state utilities of study respondents with psychological distress by age group: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S3b: Predicted mean health state utilities of study respondents with psychological distress by sex: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S4a: Predicted mean physical function scores of study respondents with psychological distress by age group: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S4b: Predicted mean physical function scores of study respondents with psychological distress by sex: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S5a: Predicted mean role physical scores of study respondents with psychological distress by age group: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S5b: Predicted mean role physical scores of study respondents with psychological distress by sex: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S6a: Predicted mean mental health scores of study respondents with psychological distress by age group: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S6b: Predicted mean mental health scores of study respondents with psychological distress by sex: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S7a: Predicted mean role emotional health scores of study respondents with psychological distress by age group: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S7b: Predicted mean role emotional health scores of study respondents with psychological distress by sex: full sample analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Fig. S1a & S1b: Mean physical function scores among Australian adults by psychological distress severity and age, and by psychological distress severity and sex, based on the Household, Income and Labour Dynamics in Australia Survey, 2007–2021. Fig. S2a & S2b: Mean role physical scores among Australian adults by psychological distress severity and age, and by psychological distress severity and sex, based on the Household, Income and Labour Dynamics in Australia Survey, 2007–2021. Fig. S3a & S3b: Mean mental health scores among Australian adults by psychological distress severity and age, and by psychological distress severity and sex, based on the Household, Income and Labour Dynamics in Australia Survey, 2007–2021. Fig. S4a & S4b: Mean role emotional scores among Australian adults by psychological distress severity and age, and by psychological distress severity and sex, based on the Household, Income and Labour Dynamics in Australia Survey, 2007–2021. Table S8a: Linear mixed effects models with predictors of health-related quality of life and its domains among respondents with complete data for all 8 waves: analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021). Table S8b: Linear mixed effects models with predictors of health-related quality of life among respondents with data for fewer than 8 waves: analysis from the Household, Income and Labour Dynamics in Australia Survey (2007–2021).