Understanding health problems in people with extremely low health-related quality of life in Korea

Little is known about patients reporting extremely poor health-related quality of life (HRQoL). This study targeted population with inferior HRQoL and examined their problems experienced with HRQoL dimensions, and impacts of different morbidities on these problems. Data were obtained from a population-based survey in Korea. HRQoL was measured by EQ-5D questionnaire and low-HRQoL population was defined as individuals whose EQ-5D utility score was among the lowest 5% of total survey population. Logistic regression models were used to evaluate the impact of fifteen morbidities on HRQoL dimensions. Of 2976 low-HRQoL participants, females and low socioeconomic individuals were predominant. They experienced significantly more problems in all dimensions, with pain/discomfort, and mobility as the most frequently reported problems. Problems in HRQoL dimensions diverged according to diseases. Individuals with arthritis experienced more difficulties with mobility (aOR 2.62, 95% CI 1.77–3.87) and pain/discomfort (aOR 2.86, 95% CI 1.78–4.60). Stroke patients experienced more problems in self-care (aOR 2.24, 95% CI 1.59–3.15) and usual activities (aOR 1.87, 95% CI 1.11–3.14). Having two or more diseases was associated with worse outcomes in usual activities and increased risk of depression. Thus, efforts to improve status of low-HRQoL should be customized to fulfil unmet needs corresponding to various diseases, and depression prevention is needed for those with multimorbidity status.

www.nature.com/scientificreports/ The EuroQoL (EQ-5D) questionnaire is one of the most widely used preference-based instruments assessing HRQoL, with five dimensions: mobility function, self-care, usual activity, pain or discomfort, and anxiety or depression. Recently, there has been increasing awareness and efforts at various levels to measure HRQoL in Korea. In Korea, EQ-5D was measured in an annual nationwide health survey, namely the Korean National Health and Nutrition Examination Survey (KNHANES) since 2005 7 . Using this population-based survey, we first targeted those who reported having extremely low QoL and aimed to understand their demographic as well as morbidity characteristics. Second, we examined how morbidities negatively impact different HRQoL dimensions. Additionally, as previous evidence yields gender and socioeconomic inequalities in HRQoL in the Korean population 8,9 , we investigated whether inequalities exist among this low-HRQoL population.

Results
Characteristics of the low-HRQoL population. A total of 50,583 participants were included in our study population. Of those, 2,976 participants were classified as a low-HRQoL population. Of 2,976 subjects reporting low QoL, 58.6% were elderly (age > 65), 68.2% were female, 41.3% had low income, 68.7% had an elementary education, and the mean age was 65.2 (0.4) years old (Table 1). Only 16.8% of the low-HRQoL population were disease-free, compared to 61.4% among the general population. In the low-HRQoL population, 47.1% had one or two diseases, and more than one-third (36.1%) had three or more. Further, in the low-HRQoL population, hypertension was the most prevalent disease (47.0%), followed by arthritis (42.5%), diabetes (21.3%), and hyperlipidemia (19.7%). Meanwhile, among the total population, the highest percentage of patients were diagnosed with hypertension (16.8%), followed by arthritis (9.8%), hyperlipidemia (8.8%), and diabetes (6.5%). When asked about overall health status in the low QoL population, 18.9% answered that they had normal overall health status, while 73.5% responded that their health status was bad or very bad.
Dimensions of HRQoL experiencing problems in low-HRQoL population versus total survey population. The low-HRQoL population experienced more problems in all HRQoL dimensions than the total survey population (Fig. 1). Proportions of reporting problems were significantly higher in the low-HRQoL group than those of the rest of the survey respondents (non-low-HRQoL, all p-values < 0.001). More problems were noted in pain/discomfort (95% versus 19%) and mobility dimension (90% versus 9%). In terms of other dimensions, 85% of low-HRQoL populations reported having problems with their usual activities, 62% reported having self-care problems, and 55% experienced anxiety or depression.

Divergence of HRQoL dimensions by morbidity.
HRQoL dimensions diverged from disease to disease among the low-HRQoL population. Mobility problems were reported most frequently for some circulatory diseases, including stroke (95.4%) and arthritis (95.6%) ( Table 2). Meanwhile, patients with hypertension, stroke, diabetes, and CKD reported more self-care problems than those without the disease. Stroke patients also reported significantly more problems in the usual activity dimension than the non-stroke population (93.3%, p < 0.001). More problems in the usual activity dimension were reported in patients with angina pectoris, asthma, and CKD. Arthritis patients experienced significantly more pain and discomfort than others (97.6%, p < 0.001, respectively).
After adjusting for confounding covariates, the multivariate logistic regression models estimate how dimensions are significantly associated with the diseases (Fig. 2). The odds of reporting problems in the HRQoL dimensions varied by disease type. Individuals with stroke reported higher odds of experiencing self-care (aOR 2.24; 95% CI 1.59 to 3.15) and usual activity (aOR 1.87; 95% CI 1.11 to 3.14) problems. Diabetes patients experienced significantly more self-care problems (aOR 1.32; 95% CI 1.06 to 1.65). Individuals with a thyroid disorder (aOR 2.57; 95% CI 1.06 to 6.22) and arthritis (aOR 2.62; 95% CI 1.77 to 3.87) experienced significantly more mobility problems. People with arthritis also showed higher odds of experiencing pain/discomfort (aOR 2.86; 95% CI 1.78 to 4.60).
The impact of multimorbidity on HRQoL among the low-HRQoL population was also evaluated (Fig. 2). Our results indicate that when participants had multimorbidity, they experienced significantly more usual activity problems: one or two disease conditions (aOR 1.60; 95% CI 1.10 to 2.31) and three or more comorbidities (aOR 1.80; 95% CI 1.20 to 2.72). Risk of having depression or anxiety also increased when participants had three or more comorbidities (aOR 1.40; 95% CI 1.02 to 1.92).

Discussion
Recently, health promotion has expanded from focusing on mortality and morbidity to incorporating interventions that enhance the community's HRQoL. This article contributes to society health literature by exploring the population who experienced extremely low HRQoL and, further, better understanding their problems in disease-comparative settings. The explanatory power of our findings is relatively high since the analysis was based on nationwide data, and it corroborates previous results for this specific population. Some unique findings were gained from both their demographic and disease-characteristic perspectives. Problems with mobility function and pain/discomfort were most frequently reported in the low-HRQoL population relative to the general population. Problems in HRQoL dimensions are also diverged by the disease. A higher number of coexisting disease conditions were associated with limitations in their usual activities and increased the risk of depression. Our results further suggest that disparities in gender and socioeconomic status existed in the low-HRQoL population.
In this study, two of every three low-HRQoL people were female and had the lowest education level. Lower socioeconomic status and female gender thus exacerbate the risks of poor HRQoL outcomes; this "social gradient" in HRQoL is also evident in the previous studies 9, 10 . Our results also indicate that the elderly population is dominant in the low-HRQoL population, accounting for approximately 80%. As age increases, the elderly might www.nature.com/scientificreports/ experience frailty, not only in their physical conditions but also in their psychological functions. Aging correlates with poorer HRQoL outcomes 8,9,11 . A notable distribution of elderly in the low-HRQoL population found in this study suggests that societal strategies to promote the status of these elderly are needed.
In the low-HRQoL population, pain and discomfort are the most frequently reported problems, consistent with previous evidence [12][13][14][15] . However, the impact on HRQoL dimensions differed by disease. Some specific patterns in how different diseases deteriorate HRQoL dimensions were explored. Stroke patients experience more problems with self-care and usual activities than the non-disease group. Meanwhile, individuals with arthritis suffered more in terms of mobility and pain/discomfort. Arthritis patients usually suffer from significant pain and disability, deteriorating their HRQoL [16][17][18] . Thyroid disorder patients experienced more difficulties in daily activity functions 19 . Thyroid disorders are rarely life-threatening; however, they can diminish patients' HRQoL because of the thyroid's essential roles 20 . www.nature.com/scientificreports/ Depressive disorders are frequently comorbid with long-standing chronic conditions such as hypertension, diabetes, and cardiovascular diseases 21,22 . An increasing number of these coexisting diseases correlate with a higher risk of depression and consequently worsen HRQoL [23][24][25] , as found in this study. Depression and anxiety might cause barriers to a treatment course for the comorbidities and therefore worsen health outcomes 26 . Even though patients with multi-comorbidity may experience lower HRQoL than others, they may not receive the attention they deserve 27,28 . Our findings highlight that individuals with multiple disease conditions, particularly depression and anxiety, should be prioritized to diminish disease burden and improve overall HRH.
The EQ-5D used in this study has national social tariffs, also called a value set, derived from the Korean population 29 . However, concerns have been raised about the ceiling effects of the EQ-5D when used in general populations 30,31 . A high proportion of the perfect scores in HRQoL dimensions have resulted in high mean EQ-5D indexes in different populations 14 . This skewed nature of the EQ-5D scores causes a lack of discrimination among mild health states when analyzing HRQoL data from KNHANES. Given this phenomenon, we aimed to target only those who reported the lowest 5% of EQ-5D utility scores in this study and considered them as the low-HRQoL population. We believed that this approach might result in better discrimination of low-and non-low-HRQoL populations.
As abovementioned, the current study aimed at the participants who reported extremely low quality of life, and thus, the average EQ-5D index score was 0.72, which is substantially lower than the average score of 0.96 of the total KNHANES study population 32 . The EQ-5D index value of 0.72 thus indicates a disparity in people with extremely low HRQoL relative to the general population. In addition, the recent statistics indicate that the EQ-5D index decreased considerably with increasing age, and a gender gap, with worse HRQoL in females, was reported. Therefore, future health care policy or welfare programs are needed to shorten the gap in HRQoL between the low HRQoL and the general population, with additional consideration of the inequalities arising from age and gender.
Two main points should be explicitly acknowledged and considered when interpreting the results of our study. First, since KNHANES is meant for non-hospitalized civilians only, persons with severe conditions were likely to be excluded. Therefore, for some diseases well known to lead to low HRQoL, such as stroke or cancer, as measured in this survey, HRQoL status might be underestimated. Other limitations of our study should also be acknowledged. For example, KNHANES only includes the 15 most common disease types, whereas different disease types or external factors might impact this low-HRQoL population.
In general, the low-HRQoL population endures the most pain, discomfort, and mobility function issues, emphasizing the need for a more comprehensive approach concentrating on these problems. In addition, those with more coexisting diseases might benefit from interventions to prevent depression. Since the impact on  www.nature.com/scientificreports/ more frequently experienced lower HRQoL than others. Hence, HRQoL inequalities arising from gender and socioeconomic status merit further investigation and appropriate interventions.

Method
Participants. Data were obtained from the multi-year KNHANES from 2007 to 2015. The KNHANES is a nationwide population-based complex multistage survey. The participants were chosen by a complex proportional allocation system and systematic sampling with multistage stratification, age, sex, and region 7 . This study was approved by the National Cancer Center Institutional Review Board of Korea (Approval Number: NCC2018-0284), and was conducted in accordance to the guidelines of the Declaration of Helsinki-ethnical principles for medical research involving human subjects. All of the participants provided informed consent prior to participating in the KNHANES.

Measures.
Health-related quality of life measure. HRQoL was measured using EuroQoL Five-Dimension (EQ-5D). The five dimensions were measured by five related questions evaluating a participant's health status: mobility, self-care, usual activity, pain/discomfort, and depression/anxiety. Response levels were as follows: "no problem," "moderate problem," and "extreme problem." In this study, we re-categorized the three classes into two groups: "report no problem" and "report problem" (either moderate or extreme problems). The Korean version of the EQ-5D was cross-culturally adapted and validated in a previous study 33,34 . The kappa value of EQ-5D dimensions between test and retest was 0.32-0.64, and the intraclass correlation coefficient of the EQ-5D index was 0.61. Therefore, EQ-5D is considered a useful instrument for measuring HRQoL of the general Korean population, with a moderate convergent and discriminant validity 33,34 .
Low-HRQoL population. The EQ-5D index score was calculated using the Korean value set for this instrument 29 .
The range of EQ-5D index scores was from −0.17 to 1. One indicates perfect health condition, zero indicates a condition as bad as death, and less than zero indicates a subjective condition worse than death 35 . We defined our target population as the lowest 5% of the EQ-5D index of the total KNHANES population (EQ-5D index score ≤ 0.72).
Morbidity status. Fifteen diseases were included in our analysis: hypertension, hyperlipidemia, stroke, myocardial infarction, angina pectoris, arthritis, pulmonary tuberculosis, asthma, cancer, diabetes mellitus, thyroid, depression, liver diseases (hepatitis B, hepatitis C, hepatocirrhosis), renal failure, and chronic kidney disease (CKD). Subjects were asked about their disease status for a range of diseases in the form of an adjustable question: for example, "Have you ever been diagnosed with stomach cancer by a doctor?" Those who answered "yes" were considered to have that disease. CKD disease was defined as a glomerular filtration rate less than 60 mL/ min/1.73m 2 , calculated from creatinine level, in the health examination 36 . Details on each disease measurement were described in Supplementary Table 1.
Demographic factors. Sociodemographic factors included age, gender, income level, and education level. Income was categorized into four quartile household income groups: low (the lowest 25%), middle-low (between 25 and 50%), middle (between 50 and 75%), and high (the highest 25%). Education levels were categorized into elementary graduate or lower, middle school graduate, high school graduate, and college graduate or higher.
Statistical analysis. Descriptive statistics were calculated to describe the participants' sociodemographic characteristics, and chi-squared tests were used to compare distributions in HRQoL problems. We fitted logistic regression models to evaluate the probability of individual reporting problems in each EQ-5D dimension by disease status. In the model, the non-disease counterparts were used as the reference group (e.g., cancer versus non-cancer). We fitted a model for each disease and adjusted for age, gender, income level, and education level. In our analysis, we considered the stratified multistage clustered probability sampling design and survey weights. All statistical analyses were performed using the SAS survey procedures in SAS software version 9.4 (SAS Inc., Cary, NC). P-values less than 0.05 were considered statistically significant.