Newly estimated disability weights for 196 health states in Hubei Province, China

Abstract

Background

The disability weight (DW) reflects the severity of non-fatal outcomes and is an important parameter in calculating the burden of disease. However, the universality of the global, national, or subnational DWs remains controversial. This study aims to measure DWs specific to Hubei Province of China using non-parametric regression to anchor the DWs.

Methods

Paired comparison (PC) data collected from a web-based survey in Hubei Province targeting the general population were used to estimate the DWs of 196 health states. Specifically, PC data from 33,925 respondents were analyzed by probit regression analysis, and the results were then anchored to 0–1 scale using non-parametric regression based on the DWs from Global Burden of Disease (GBD) 2013. The absolute DW values and rankings were compared to those in the Chinese disability weight measurement study, GBD 2013, and Japan.

Results

The DWs for 196 health states ranged from 0.003 for mild distance vision impairment to 0.663 for severe heroin and opioid dependence in Hubei Province, China. Quite a lot mental disorders, such as moderate/severe episode of major depressive disorder, were considered more severe than the terminal phase with/without medication among Hubei residents. DW rankings of the health states are relatively stable in Hubei Province irrespective of the anchoring method used. A very small proportion (4 of 196, 2%) of DW rankings changed by 10 or more positions in China when compared with our results, but approximately 61% in GBD 2013 and 59% in Japan. Among the top 25 health states in this study, 9 of 11 health states categorized as mental, behavioral, and substance use disorders resulted in a lower ranking in GBD 2013, and all 6 states in Japan also showed a lower ranking, whereas China shared a similar ranking.

Conclusions

The burden of mental disorders among Hubei residents, especially moderate or severe major depressive disorder, deserves further attention. When using different anchoring methods, DW rankings were maintained relatively stable rather than the absolute values in Hubei. Substantial differences of DW rankings between our results and that in China, GBD 2013, and Japan draw attention to the need for deriving local disability for disease burden calculation.

Background

Injury and disease can lead to many disabling conditions that have a substantial negative impact on the quality of life. Understanding and assessing the loss of health associated with these conditions are vital for the calculation of disease burden and effective health policy-making, especially in developing countries that generally have multiple competing demands at the same time [1]. Disability-adjusted life years (DALYs) is a summary measure reflecting disease burden. It combines years lost due to premature mortality with the loss experienced by living with a disability, that is, the sum of years of life lost (YLLs) and years lived with disability (YLDs) [2,3,4]. Thereinto, to calculate YLDs, disability weight (DW) is an essential parameter [5, 6], which is a numerical scale between 0 and 1 and represents the severity of the health conditions, with 0 being full health and 1 being death [7, 8].

There have been some studies that established DWs for a wide range of diseases and injuries, the most comprehensive and systematic of which was the Global Burden of Disease (GBD) studies that can be traced back to the groundbreaking research in 1996 [9]. Subsequently, a handful of related studies emerged or were updated, with conceptual, methodological, and technical modifications [10,11,12,13,14]. Certainly, up to now, the most widely used approach for the computation of DWs belongs to that described in the GBD 2010 study [15, 16], and perhaps the most extensive survey is GBD 2013 which covered 235 health states and involved 60,890 subjects from 9 countries [7, 8]. Unfortunately, respondents from Africa or Asia (like China and India) are rather limited, and most come from America or Western European countries, which raises questions about the universality of DWs reported in GBD studies [17]. Although, previous research has shown relatively stable rankings of health conditions across 14 countries [18]. A recently published study that estimated the DWs for 231 health states based on 37,318 Japanese respondents has provided supporting evidence for the variability between DWs in the East and West [19]. Furthermore, studies from China and its provinces also found a certain difference across countries or provinces [20, 21]. Recent studies in Wuhan, China have also suggested the necessity of estimating local DWs [22]. Although the DWs of the Hubei-Provincial level have been reported in the Chinese disability weight measurement study, it is based on paired comparison (PC) responses from Hubei Province and population health equivalence (PHE) responses from all over the country. Previous research suggests that different anchoring methods may give different results [23]. Moreover, data on how methodological differences can impact absolute DW value and its rankings are limited. Therefore, it is quite necessary to understand the differences resulting from anchoring methods to better evaluate the perceptions of health states among Hubei residents and how they differ from other populations.

This study aimed to estimate the DWs of 196 health states specific to Hubei Province using an anchoring procedure based on the results from GBD 2013 and examine the differences between them and the Hubei-provincial results from the Chinese disability weight measurement study which anchoring procedure based on the results from population health equivalence (PHE) responses of Chinese. Additionally, we compare our results with the national results of China, GBD 2013, and Japan to explore the cultural or geographical boundaries of DWs between regions or countries to better assess the disease-specific burden.

Methods

Study design and participants

This research was implemented as part of the Chinese disability weight measurement study [20], which was designed based on the online survey protocol applied in the European disability weights measurement study and GBD studies [8, 17]. Put simply, this study was conducted through a secure, web-based platform between May 12, 2020, and July 22, 2020, in Hubei province, and participants were recruited from all 17 cities with a stratified random sampling by various channels such as working group meetings of the investigation team, community publicity, online social network, and word of mouth referrals. All web-based surveys were administered via WeChat, with the data captured and transmitted online. All participants were between the age of 18 and 69 years.

Description of health states and valuation method

We estimated DWs for 196 health states which reflected the severity of non-fatal outcomes caused by disease or injury to a patient’s physical health and psychological and social functioning. Each health state was described by a brief lay description highlighting major functional consequences and symptoms. Additional file 2 showed lay descriptions for 196 health states of the present study.

The online questionnaire was comprised of sociodemographic and 16 PC questions originating from a pool of lay descriptions [20]. Paired comparison is an ordinal measurement method that obtains a priority scale through binary comparisons. With this method, participants were asked to choose the healthier option between two random pairs of health states. Specifically, questions 3, 10, and 16 were designed to overlap with other questions as quality control.

Statistical analysis

Most analyses and procedures were the same as in the Chinese disability weight measurement study [20]. The difference is that this study rescaled the results of probit regression analysis using the DWs of GBD 2013 by the rescaling procedure used in the European disability weights measurement study [9]. Firstly, after eliminating participants who failed quality control, the PC data was included in a probit regression analysis where the response variable is the choice responses of 0 or 1 (where 1 indicates that the first state is selected as the healthier outcome and 0 otherwise) and the indicator variables are each health state (that the first state in a PC is assigned a value of 1, the second state in a PC is assigned a value of -1, and 0 otherwise). Secondly, locally weighted linear regression (Loess) was applied to fit the probit regression coefficients against the 172 corresponding logit-transformed DWs from GBD 2013. Loess combines the simplicity of linear least squares regression with the flexibility of nonlinear regression. and fits simple models to localized subsets of the data to build up a function that describes the variation in the data, point by point. Consequently, predicted logit-transformed DWs for each probit coefficient were obtained based on the fitted Loess curve. Lastly, each predicted value was transformed to the natural 0–1 scale through an inverse logit function. Furthermore, the uncertainty interval (UI) of each DW was estimated by Monte Carlo simulations with 1000 iterations. Specifically, 1000 random probit regression coefficients derived from the probit estimated mean and standard deviation for each of the DWs were used to repeatedly fit Loess and calculate DWs, and 95% UIs were subsequently calculated with the 2.5th and 97.5th percentiles of the distribution of each DW. In addition, Pearson correlation coefficients were used to assess changes in probit regression coefficients for age, sex, educational and household income level, and medical background. Pearson correlation analysis was also used to determine the correlation between the DW values of this study and those reported in the Chinese disability weight measurement study [20], GBD 2013 study [8], and Japanese study [19]. A P-value less than 0.05 was considered statistically significant.

All data preparation and statistical analyses were performed using SAS (version 9.4; SAS Institute, Inc.) and R, version 4.1.2.

Results

The characteristics of respondents

A total of 33,925 participants met our inclusion criteria and were included in the final analysis. Overall, the participation occurred in all 17 cities of Hubei Province, 10 of which had at least 2,000 participants. The general characteristics of study participants are summarized in Additional File 1. The median age of the participants was 42 years (interquartile range [IQR], 32–48 years) and 63.8% were women. Furthermore, the majority of participants self-identified as having a background in medicine (71.5%), with a college education or above (64.1%), and reported an annual household income of more than 50,000 yuan (66.7%). The smooth transition in the heat map substantiated a comparatively small quantity of measurement error and high internal consistency in the web survey (Additional file 1). The correlations of probit regression coefficients were relatively high between different age groups ($\:{r}_{s}$ range: 0.973–0.991), sexes ($\:{r}_{s}$=0.993), education levels ($\:{r}_{s}$ range: 0.992–0.996), household income levels ($\:{r}_{s}$ range: 0.982–0.995), and medical background ($\:{r}_{s}$=0.979). All p-values were less than 0.001.

Estimated disability weights

The estimated DWs and 95% UIs for 196 health states in Hubei Province are shown in Additional File 1. Across all states, most estimated weights resided near the mild end of the severity on a scale from 0 (full health) to 1 (death), and 54 (27.55%) of the states had DWs below 0.05 (Fig. 1). Specifically, severe heroin and other opioid dependence had the highest DW of 0.663 (95%UI 0.527–0.765), followed by acute schizophrenia, sitting at 0.657 (0.526–0.763), whereas mild distance vision impairment had the lowest DW of 0.003 (0.002–0.008), followed by mild anemia, sitting at 0.004 (0.002–0.008). Severe episode of major depressive disorder and end-stage renal disease (on dialysis) ranked high at 3rd and 4th, respectively, which indicated high severity of disability. Similarly, moderate episode of major depressive disorder and terminal phase with or without medication (for cancers and end-stage kidney or liver disease) were also more likely to result in high levels of disability, as they ranked high at 14th, 15th, and 21st, respectively. The results illustrated that people living in Hubei believed that severe and moderate episode of major depressive disorder tend to result in greater health losses than the terminal phase (for cancers and end-stage kidney or liver disease). Moreover, only five states of the results seemed counterintuitive for the health states with a logical order of severity level, including those related to the terminal phase of cancer with/without medication, hearing loss, distance vision, and neck pain.

Fig. 1

Distribution patterns of the disability weights in this study (Hubei) and Hubei-provincial results from the Chinese disability weight measurement study (Hubei^a), the national results from the Chinese disability weight measurement study (China), the GBD 2013 study (GBD 2013), and the Japanese study (Japan)GBD=Global Burden of Disease

In comparison with the Chinese disability weight measurement study, GBD 2013 and Japan

The disability weights of this study were highly correlated with the Hubei-provincial and national results in the Chinese disability weight measurement study (The Pearson correlation $\:{r}_{s}$: 0.993 and 0.990, respectively, P < 0.001), which are based on the results from the PHE responses. However, except for somatoform disorder and post-acute consequences (fatigue, emotional lability, and insomnia) of infectious disease, the absolute DW value of Hubei Province in this study varied to some extent from those in the Chinese study. As shown in Fig. 2, a higher DW was estimated in this study when the value ranges from approximately 0.1 to 0.5. On the contrary, it was lower, when the value was out of the range. Nevertheless, the DW rankings of 196 states of Hubei Province in both studies were almost identical with only small differences (Fig. 3). Compared to the national results, this study found that the DW ranking of Crohn’s disease or ulcerative colitis moved up ten places and ranked at 82nd, whereas invasive device/drain, herpes zoster, and local lymphogranuloma venereum infection moved down equal to or more than ten places and ranked at 87th, 172nd, and 120th, respectively (Fig. 3). Furthermore, the ranking changes for the top and the bottom 25 health states were relatively small with no more than 5 positions when compared to the national results (Figs. 4 and 5).

As shown in Fig. 1 and Additional file 1, the shape of the DW distributions was right-skewed in all studies and relatively highly correlated ($\:{r}_{s}$≥0.84, P < 0.001). However, considerable heterogeneity in the ranking changes between the DWs of this study and those of GBD 2013 and the Japanese study was also observed. In comparison to GBD 2013, this study found 121 DWs rankings changed by 10 or more positions, and a change in rank of greater than 50 places for health states related to mild cannabis dependence, mild amphetamine dependence, mild cocaine dependence, tension-type headache, most severe neck pain, bulimia nervosa and moderate motor plus cognitive impairments, the first four of which had 9.56 times, 4.11 times, 2.56 times, and 4.59 times DWs than GBD 2013, respectively (Additional file 2 and Fig. 3). In addition, among the health states of the top 25 DWs in Hubei Province, 9 of 11 states that were categorized as mental, behavioral, and substance use disorders had higher ranks than that in GBD 2013 (Fig. 6), whereas all four mental, behavioral, and substance use disorders among the bottom 25 health states had lower ranks (Fig. 7). Moreover, this study found lower DWs and ranks for health states mentioning alcohol use compared with GBD 2013. Compared to the Japanese study, this study found that 59% of DWs rankings changed by 10 or more positions, and the rank of severe anemia moved up more than 70 places, whereas bulimia nervosa and herpes zoster moved down more than 60 places according to the reordering of DWs of 165 health states that are common to this study and Japan. But again, DWs and ranks of the health states mentioning alcohol use were lower in this study (Additional file 2).

Fig. 2

Differences of the disability weights in this study (Hubei) and Hubei-provincial results from the Chinese disability weight measurement study (Hubei^a)

Fig. 3

Ranking changes of the disability weights in this study (Hubei) and those in the Chinese disability weight measurement study (Hubei^a and China), the GBD 2013 study (GBD 2013), and the Japanese study (Japan). (The left two scatterplots use the left y axis, whereas the right two plots use the right secondary y axis; Ranking changes was calculated using the ranking in Hubei minus the ranking in other studies)GBD=Global Burden of Disease

Fig. 4

Ranking of the top 25 DWs of 196 health states in both this study and the Chinese disability weight measurement study (Hubei^a means Hubei-provincial results from the Chinese disability weight measurement study). DW=Disability Weight

Fig. 5

Ranking of the last 25 DWs of 196 health states in both this study and the Chinese disability weight measurement study. DW=Disability Weight

Fig. 6

Ranking of the top 25 DWs of 196 health states in both this study and GBD 2013. GBD=Global Burden of Disease, DW=Disability Weight

Fig. 7

Ranking of the last 25 DWs of 196 health states in both this study and GBD 2013. GBD=Global Burden of Disease, DW=Disability Weight

Discussion

As a sub-study of the Chinese disability weight measurement study [21], we calculated the DWs of Hubei Province for a broad array of health outcomes by an anchoring procedure based on the results from GBD 2013 instead of PHE responses that were adopted in the Chinese disability weight measurement study, GBD 2013 and Japanese study. The range of Hubei-provincial DW values of 196 health states from this study (from 0.003 for distance vision mild impairment to 0.663 for severe heroin and other opioid dependence) was lower than the Hubei-Provincial levels in the Chinese study (from 0.010 for distance vision mild impairment to 0.708 for severe heroin and other opioid dependence) and a higher DW was estimated in this study when the value ranges from approximately 0.1 to 0.5, but about 97% DWs held the same ranking or changed only one position in both studies.

It means that the method or data used for the anchoring process did affect the absolute DW values to a certain degree but not the DW rankings. This may provide insightful evidence for using the comparison of DW rankings between different studies to better understand variations across geographical regions or countries that may result from contextual differences. Previous studies have shown that the DW rankings of the health states or disease stage are relatively stable across countries irrespective of the method used [19, 23]. Despite this, some pronounced differences, such as the difference in the ranking of drug dependence between China and other countries indicated possible cultural differences [19]. This study certainly found more obvious differences between the DW rankings of Hubei Province and those of China, GBD 2013, and Japan even though their absolute DW values were highly correlated ($\:{r}_{s}$≥0.86, P < 0.001).

The health state ranked most variably between Hubei Province and China was Invasive device/drain, followed by Herpes zoster, Local Lymphogranuloma Venereum infection, and Crohn’s disease or ulcerative colitis, all of which moved ten or more places. In the light of the identical description and the way questions were asked of the health state, we suggested that these changes can be mainly attributed to the difference in perceptions of state severity. On the one hand, a short lack of disease label description of the state makes respondents more susceptible to their knowledge of medicine. On the other hand, for certain autoimmune chronic diseases that recur easily (e.g. Crohn’s disease or ulcerative colitis), differences in accessibility and convenience of obtaining medication or treatment services may cause heterogeneity in the severity of DW. Some studies have shown that depression and anxiety became major problems affecting patients with inflammatory bowel disease in Hubei Province during the COVID-19 epidemic when this survey was conducted [24, 25]. Therefore, understanding the context of deriving disability weights is important for exploring reasons for differences.

Moreover, it is worth noting that manic episode of bipolar disorder and moderate episode of major depressive disorder were considered more serious than the terminal phase with medication (for cancers and end-stage kidney or liver disease) in Hubei Province, but the result was reversed in China, GBD 2013 and Japan [8, 20, 21]. Nevertheless, unlike in the GBD 2013 and Japan, the ranks of the three states mentioned above were adjacent in China. It highlights the severe loss of health caused by mental disorders mentioned above in China, especially in Hubei Province. This may be linked to the high prevalence of stigma and discrimination against people with mental illness in China [26, 27]. Research has also shown that quite a few people have mental health problems in Hubei Province during the COVID-19 outbreak [28], and some young people are depressed during the post-COVID-19 epidemic period [29]. Hence, it is highly necessary to develop policies to prevent mental disorders and provide more assistance and public attention to people with such disorders in Hubei Province.

The state with the most variation of ranking between this study and GBD 2013 was mild cannabis dependence, followed by mild amphetamine dependence, both of which had significantly risen in rank in Hubei Province. Conversely, bulimia nervosa experienced the most significant drop in rank in Hubei Province. A previous study has also noted a fairly high ranking for drug dependence in China [19]. Similarly, mild drug dependence also ranks very highly in Japan [20]. However, unlike the present study, bulimia nervosa was assigned considerably higher DW values and ranking in both the Japanese and South Korea surveys [20, 30], which is consistent with the results from GBD 2013. Furthermore, we also found a lower ranking of alcohol use disorder-related health states in this study compared to GBD 2013 and Japan.

Together these results may indicate a cultural difference in how certain health outcomes are viewed in different regions. For instance, poor knowledge, negative attitudes toward drug dependence, and limited availability of information on and treatment for drug dependence create a vicious cycle of social rejection, marginalization, and relapse for drug abusers in China [31, 32]. Drug-related crime accounts for 70–80% of all crimes in some areas of China [33]. Similarly, in Japan, possession of drugs is considered a serious crime, and severe penalties would be given for the offense [34, 35]. However, European countries, such as the United States and Germany, have less harsh penalties for drug possession and relatively perfect social-support systems [32,33,34,35,36,37]. At the same time, mental health professionals in Western countries like Australia have positive attitudes toward drugs and substance abuse [38]. These differences could be a reason for the variation in the ranking of drug dependence in different regions.

As another example, the lower ranking of bulimia nervosa in both this study and the Chinese surveys may be related to inadequate information about bulimia nervosa and a lack of culturally appropriate bulimia nervosa information [39, 40]. Additionally, in those cases when alcohol use disorder-related states have much lower ranks, it might be owing to the Chinese culture of drinking, which is a well-accepted means of relieving physical and psychological symptoms [41]. In all, there are still quite a few differences across regions, especially between different countries. We found that 11 health states with the top 25 DWs in Hubei Province classify as mental, behavioral, and substance use disorders and 9 of them had higher rankings than GBD 2013. Besides, 6 of 25 states common to Japan were also classified as mental, behavioral, and substance use disorders and had higher rankings in Hubei Province. Apart from cultural differences, general explanations might cover different ethnicities, different disease states, or self-cognition [42,43,44,45,46]. This further illustrates the importance of deriving local DWs.

Finally, it should be pointed out that the severity-disaggregated DWs for certain health states seem counterintuitive in this study. For instance, the DW for terminal phase of cancer with medication was higher than that without medication, and the DW for severe neck pain was higher than that of most severe neck pain. In the case of “distance vision impairment”, the DWs were 0.241 for severe impairment, but 0.230 for blindness. Similar situations were also observed in “hearing loss” and “hearing loss, with ringing”. On the one hand, these inconsistencies could be explained by the Chinese lay descriptions used in the survey. It is possible that respondents might misinterpreted some aspects of the lay descriptions, or the relative consequences of certain states might have been improperly described. For example, terminal phase without medication only “has constant pain”, but terminal phase with medication has “regularly uses strong medication to avoid constant pain” included in the description which may make respondents feel worse. On the other hand, the sample population used in the survey does not fully represent the vision/hearing loss population, which has little access to healthcare services especially in rural areas. Therefore, it is worth considering the possibility of different results from specific populations in future research. Certainly, more quality control questions are necessary to be introduced to reduce the impact of respondents who give illogical answers just like the recent research in Korea [47].

Despite important implications, some limitations of this study are worthwhile to consider. First, this web-based survey does not cover participants younger than 18 years and older than 69 years and may have an under-representation of younger or elderly people because of the possibly different perceptions of health states between younger and older persons [48]. But, that simultaneously reduces the introduction of errors that may be caused by not-proficient skills on the internet or lack of comprehension of questions. Also, a higher proportion of participants were women and had medical backgrounds, higher educational and income levels, but the correlations of probit regression coefficients between different subgroups were high (> 0.973). Despite this, it must be taken into account that variation of the characteristics might have an impact on the final derived DWs, and further research is needed in the future. Second, although stringent quality control measures have been taken and a web-based survey is quite economical, face-to-face interviews can result in higher quality data and will be a consideration for future research. Third, while a great deal of effort has been made in this research to attenuate the discrepancy between the English and Chinese languages, some variability remained and perhaps contributed to part of the differences in DW rankings.

Conclusions

In conclusion, the distributions of DW values in different countries or regions are right-skewed. The different anchoring processes may affect to some extent the absolute DW values, but the rankings of DW are relatively stable. Therefore, comparing the DW rankings of different countries in different studies may be a better choice to solve the difference caused by methods, even if there is no gold standard. In this study, we found that the DW ranking of moderate episode of major depressive disorder is higher than the terminal phase with/without medication (for cancers and end-stage kidney or liver disease), which is different from the Chinese, GBD 2013 and Japanese survey. Moreover, most mental disorders of the top 25 DWs are considered more severe among Hubei residents as compared with GBD 2013 and Japan. These differences further confirmed the necessity of deriving local DWs and helped to highlight the burden of mental disorders in Hubei Province.