The impact of rheumatoid arthritis on the burden of disease in urban China

Abstract

Objectives:

The aim of this study is to assess the burden of disease associated with the impact of rheumatoid arthritis in urban China. Burden of disease is considered from four perspectives: (i) health-related quality-of-life (HRQoL); (ii) health status; (iii) employment status; and (iv) absenteeism and presenteeism.

Methods:

Data are from the 2009 National Health and Wellness Survey (NHWS) of urban China. This is an internet-based survey and details the health experience of 13,007 respondents. The survey is representative of the urban China population at 18 years of age and over (18.1% of the total population). Of those responding to the survey, a total of 353 reported that they had been diagnosed with rheumatoid arthritis – an unweighted estimate of 2.65%. The sample design allows a comparison of those reporting rheumatoid arthritis with those not reporting this disease and, hence, a quantitative assessment of the burden of disease. Estimates of the quantitative impact of the presence of rheumatoid arthritis are through a series of generalized linear regression models. HRQoL is evaluated through the SF-12 instrument together with responses to the first item of the SF-12, self-reported health status. The SF-12 instrument generates three measures of HRQoL: the physical component summary (PCS), the mental component summary (MCS) and SF-6D utilities. Health status is captured as a self-report on a 5-point scale. Employment status is considered in terms of self-reported labor force participation, while absenteeism and presenteeism are estimated from the Work Productivity Activity Index (WPAI). Apart from a binary variable capturing the presence or absence of rheumatoid arthritis, control variables were included to capture the impact of other potential determinants of HRQoL and health status.

Results:

The presence of rheumatoid arthritis in urban China has a significant deficit impact on HRQoL as measured by the PCS and MCS components of the SF-12, SF-6D absolute utilities and on self-assessed health status. In the case of PCS, the deficit impact of rheumatoid arthritis is −2.289 (95%CI: −3.042 to −1.536); for MCS −1.472 (95%CI: −2.338 to −0.605) and for utilities −0.025 (95% CI: −0.036 to −0.014). In the case of health status the odds ratio for the presence of rheumatoid arthritis is 1.275 (95%CI 1.031–1.576). The presence of rheumatoid arthritis has a marked negative effect, just under 8%, on the likelihood of workforce participation. Finally, the presence of rheumatoid arthritis is associated with an increased likelihood of absenteeism and presenteeism.

Limitations:

The NHWS survey has a number of limitations. As the NHWS is an internet-based survey, biases may be present due to the lack of internet penetration in the urban China population. The extent to which individuals and households have internet access is unknown. In addition, the NHWS relies upon respondents reporting they have been diagnosed with one or more specific disease states. These are not, given the nature of the survey, clinically verified. This also introduces a degree of uncertainty. Care should be taken in uncritically generalizing these results to the wider China population.

Conclusions:

The burden of disease associated with self-reported, diagnosed rheumatoid arthritis in urban China is substantial. Utilizing a series of multivariate models, substantial deficits are associated not only in reported HRQoL and health status but also in respect of employment status and, for those in employment, rates of absenteeism and presenteeism.

Keywords::

• China
• Rheumatoid arthritis
• HRQoL
• Employment

Introduction

The burden of rheumatoid arthritis in urban China is unknown. The aim of this study is to assess the burden of illness associated with this disease state in terms of (i) health-related quality-of-life (HRQoL); (ii) self-reported health status; (iii) employment status; and (iv) absenteeism and presenteeism. This is the first time such an analysis has been undertaken from the perspective of the impact of this disease on individuals.

Even so, rheumatoid arthritis has not been neglected. There have been a number of studies that have assessed Chinese versions of both generic and disease-specific patient reported outcomes instruments. The Chinese version of the SF-36 instrument has been evaluated by Koh et al.1 Chu et al.2 have assessed the Chinese version of the Arthritis Impact Measurement Scale, while Koh et al.3 have evaluated the Chinese version of the Health Assessment Questionnaire (HAQ) in rheumatoid arthritis. At the same time there have been a number of studies that have considered the role of traditional Chinese medicine in rheumatoid arthritis and aerobic fitness4–6.

Apart from these studies, however, there have been none that have considered the impact of rheumatoid arthritis in purely quantitative terms. There are no studies that have considered the impact of rheumatoid arthritis on employment status in urban China or, among the employed, on absenteeism and presenteeism. As such, this study adds to our understanding of the impact of rheumatoid arthritis in the population of urban China and the burden of disease in this population.

Methods

This study utilizes a series of generalized linear regression models with data from the 2009 National Health and Wellness Survey of urban China. The techniques employed have been used widely to assess the burden of disease with data from this survey and provide a robust framework for quantitative assessments7,8.

National Health and Wellness Survey

The National Health and Wellness Survey (NHWS) is a syndicated, annual, internet-based, cross-sectional study of the healthcare attitudes, behaviors, and characteristics of the adult population in eight countries including the US, the UK, France, Spain, Germany, Italy, Japan, and urban China. Since its initiation in 1998, over 1,000,000 survey responses across ∼140 conditions have been collected.

The present analysis is based on the results of the urban China NHWS for 2009 (representing 18.1% of the total adult China population). In the 2009 NHWS for China, the study was fielded to adults being representative of the urban population of China, consisting of Tier 1/Tier 2 cities. Approximately 90% of the 2009 NHWS sample come from the following Tier 1/Tier II cities: Beijing, Shanghai, Guangzhou, Changchun, Changsha, Chengdu, Cheungsha, Chongqing, Dalian, Dongguan, Fuzhou, Guiyang, Haikou, Hangzhou, Harbin, Hefei, Huheaote, Jinan, Kunming, Lanzhou, Lasa, Nanchang, Nanjing, Nanning, Ningbo, Qingdao, Shenyang, Shenzhen, Shijiazhuang, Suzhou, Taiyuan, Tianjin, Urumqi, Wuhan, Xiamen, Xi’an, Xining, Yinchuan, Zhengzhou, and Zhuhai. Additionally, there are ∼400 more cities that comprise ∼10% of the sample from the 2009 China – NHWS database. Sampling was completed using a mixed methodology where the internet was used to provide a self-administered survey to the 18–49 year old respondents, and Computer Assisted Web Interviews (CAWI) was utilized for the fielding of 50+year old respondents. Recruitment for CAWI was conducted by phone or interview and, if qualified, respondents were asked to come to a facility, which included a trained professional, and complete the survey on a laptop.

A total of 13,307 persons 18 years of age and over were interviewed. Overall, 353 indicated they had been diagnosed with rheumatoid arthritis. No attempt is made to include those only with rheumatoid arthritis; all comorbidities are implicit in the response. The unweighted self-reported prevalence of rheumatoid arthritis is 2.65% (weighted prevalence is 3.05%). These estimates are slightly higher than prevalence estimates for the US and previous estimates for China which are in the range 0.5–2.0%9,10. As such, care should be taken in interpreting these prevalence data as necessarily representative (given the exigencies of collecting such data in this environment).

Dependent variables

The instrument selected to measure HRQoL in the NHWS is the SF-12 (version 2)11. The SF-12 is a generic, multipurpose short-form with 12 questions. These questions are all selected from the SF-36 health survey12,13. The focus here is on the two summary scores that can be generated from the respective item responses. These are (i) the physical component summary (PCS) and (ii) mental component summary (MCS). Details of the scoring algorithms are given in Ware et al.11

The distribution of PCS and MCS scores for the rheumatoid arthritis and control samples in urban China are presented in Table 1. Persons diagnosed with rheumatoid arthritis have consistently lower PCS and MCS than the control population. In the case of the PCS measure, the deficit is over 5-points; for the MCS it is somewhat less.

Table 1.  Distribution by number and percentage of physical and mental component SF-12 summary scores, urban China, NHWS 2009.

Scores	PCS No. (%) Rheumatoid Arthritis	MCS No. (%) Rheumatoid Arthritis	PCS No. (%) Control population	MCS No. (%) Control population
0–9	0 (0.00)	0 (0.00)	1 (0.01)	0 (0.00)
10–19	3 (0.85)	1 (0.28)	21 (0.16)	20 (0.15)
20–29	14 (3.97)	15 (4.25)	169 (1.30)	448 (3.46)
30–39	87 (24.65)	114 (32.29)	1532 (11.83)	2562 (19.78)
40–49	178 (50.42)	149 (42.21)	4876 (37.64)	5747 (44.36)
50–59	71 (20.11)	66 (18.70)	6049 (46.70)	3662 (28.27)
60–69	0 (0.00)	8 (2.27)	304 (2.35)	511 (3.94)
70–79	0 (0.00)	0 (0.00)	2 (0.02)	4 (0.03)
80–89	0 (0.00)	0 (0.00)	0 (0.00)	0 (0.00)
90–99	0 (0.00)	0 (0.00)	0 (0.00)	0 (0.00)
100	0 (0.00)	0 (0.00)	0 (0.00)	0 (0.00)
Average (SD)	43.59 (7.51)	43.02 (8.02)	48.72 (7.43)	45.97 (8.27)

SD = standard deviation. Source: NHWS 2009.

As well as generating profile and summary PCS and MCS scores, the SF-12 can also be used to generate health state utilities. This is achieved through application of the SF-6D. The SF-6D provides a means for using the SF-36 and SF-12 in economic evaluation by estimating a preference-based single index measure for health from these data using general population values14,15. Any patient who completes the SF-36 or the SF-12 can be uniquely classified according to the SF-6D. The index has interval scoring properties and yields summary scores on a 0–1 scale. The preference weights used here, which have recently been revised, are those for a UK population16. The validity of the SF-6D in a Chinese population has been established, although a Chinese scoring algorithm for the SF-6D has not been published17,18. Distribution of SF-6D utilities is given in Table 2.

Table 2.  Distribution of OF SF-6D absolute utility scores, urban China, NHWS 2009.

SF-6D utility score	Rheumatoid Arthritis (%)	Control population (%)
0–0.9	0 (0.00)	0 (0.00)
0.1–0.19	0 (0.00)	0 (0.00)
0.2–0.19	0 (0.00)	0 (0.00)
0.3–0.39	0 (0.00)	5 (0.04)
0.4–0.49	7 (1.98)	56 (0.43)
0.50–0.59	54 (15.30)	860 (6.64)
0.60–0.69	232 (65.72)	7437 (57.41)
0.70–0.79	34 (9.63)	2018 (15.58)
0.80–0.89	22 (6.23)	1718 (13.26)
0.90–1.00	4 (1.13)	860 (6.64)
Average (SD)	0.652 (0.082)	0.702 (0.102)

SD = standard deviation. Source: NHWS 2009.

The average SF-6D score is lower for the rheumatoid arthritis population than for the controls (0.652 vs 0.702). The distribution, as expected, is skewed towards lower utility scores by category for the rheumatoid arthritis population.

Health status

The determinants of self-rated health status have been the focus of a large number of population health studies in the last 20 years. In the US, this been driven in large part by data available from the annual Behavioral Risk Factor Surveillance System (BRFSS) surveys where self-reported health status is a standard question19. Determinants of self-rated health have also been the focus of a number of studies from Japan, the European Union, Canada, and Australia20–23.

Health status responses for urban China respondents are detailed in Table 3. Compared to the control population, the distribution is skewed towards a less favorable assessment of health status in the rheumatoid arthritis population. For example, while 27.48% of the non-rheumatoid arthritis group would describe their health status as very good, the corresponding figure for the rheumatoid arthritis population is 11.33%. Also, while 58.92% of the rheumatoid arthritis population would describe their health status as fair, the corresponding figure for the non-rheumatoid arthritis population is 34.82%.

Table 3.  Self-reported health status, urban China, NHWS 2009.

Health status	Rheumatoid Arthritis (%)	Control population (%)
Excellent	7 (1.98)	620 (4.79)
Very Good	40 (11.33)	3560 (27.48)
Good	82 (23.23)	3869 (29.87)
Fair	208 (58.92)	4511 (34.82)
Poor	16 (4.53)	394 (3.04)
Total	353 (100.00)	12,954 (100.00)

Source: NHWS 2009.

Labor force status

The 2009 NHWS asks respondents their current workforce status and whether they are seeking work. Estimates of labor force status for urban China are detailed in Table 4. Persons with rheumatoid arthritis report a lower rate of labor force participation than those in the control group (64.9% vs 71.3%).

Table 4.  Labor force status, urban China, NHWS 2009.

Labor force status	Rheumatoid Arthritis (%)	Control population (%)
Employed full-time	209 (59.21)	8035 (62.03)
Employed part-time	7 (1.98)	459 (3.54)
Self-employed	8 (2.27)	492 (3.80)
Unemployed	5 (1.42)	255 (1.97)
In the labor force	229 (64.88)	9241 (71.34)
Not in the labor force	124 (35.13)	3713 (28.66)
Total	353 (100.00)	12,954 (100.00)

Source: NHWS 2009.

Absenteeism and presenteeism

The NHWS 2009 uses the Workplace Productivity and Activity Impairment Scale (WPAI) to measure the impact of health status on employment-related activities. The WPAI questionnaire measures work time missed and work and activity impairment because of a specified health problem during the past 7 days24. The validity and accuracy of the instrument has been established in a number of disease states (e.g., irritable bowel syndrome, asthma, dermatitis, Crohn’s disease)25,26.

The WPAI absenteeism and workplace questions are only relevant to those in employment – where the NHWS identifies persons who are currently employed full-time, employed part-time, or self-employed. Respondents are asked to indicate:

• During the past 7 days, how many hours did you miss from work because of your health problems? (Range 0–112 hours);

• During the past 7 days, how many hours did you miss from work because of any other reason such as vacation, holidays, time off to participate in this study? (Range 0–112 hours);

• During the past seven days, how many hours did you actually work? (Range 0–112 hours);

• During the past 7 days how much did your health problems affect your productivity while you were working? (Response on a 0–10 scale from ‘health problems had no effect on work’ to ‘health problems completely prevented me from working’); and

• During the past 7 days, how much did your health problems affect your ability to do your regular daily activities, other than work at a job? (Response on a 0–10 scale from ‘health problems had no effect on my daily activities’ to ‘health problems completely prevented me from doing my daily activities’).

Two measures of employment and activity impact are generated from these responses. These are:

1. Absenteeism: percentage of work time missed in the past 7 days; and

2. Presenteeism: extent to which productivity at work was impaired.

The extent of absenteeism and presenteeism could be further disaggregated in terms of their impact on the full-time employed, part-time employed, and self employed. In the present analysis, the focus is on all those in employment, irrespective of employment status.

Table 5 details the impact of rheumatoid arthritis on absenteeism. Over 37% of those with rheumatoid arthritis report that it affects their work, compared to less than 20% of controls. The impact of rheumatoid arthritis on presenteeism is more marked (Table 5) with only 11.61% reporting it had no effect compared to 30.73% of controls.

Table 5.  Workplace productivity and activity impairment scale: Impact of health problems on absenteeism in the past 7 days, urban China, NHWS 2009.

Absenteeism: Percentage of worktime lost	Rheumatoid Arthritis (%)	Control population (%)
0–9 (Health problems had no effect on my work)	138 (62.16)	7187 (82.06)
10–19	42 (18.92)	755 (8.62)
20–29	26 (11.71)	390 (4.45)
30–39	9 (4.05)	142 (1.62)
40–49	5 (2.25)	80 (0.91)
50–59	2 (0.90)	109 (1.24)
60–69	0 (0.00)	37 (0.42)
70–79	0 (0.00)	16 (0.18)
80–89	0 (0.00)	9 (0.10)
90–99	0 (0.00)	5 (0.06)
Health problems completely prevented me from working	0 (0.00)	28 (0.32)

Source: NHWS 2009.

Independent variables

Independent variables included in the model include:

• respondent socio-demographic characteristics (age, gender, household income, employment status, and education),

• respondent health risk behaviors (BMI, alcohol use, smoking), and

• presence of comorbidities (Charlson co-morbidity index).

The relationship between educational attainment, HRQoL, and healthcare resource utilization is less well established, although the expectation is that persons with higher levels of educational attainment would report greater HRQoL and a better assessed health status. Educational attainment and its association with income may also be expected to result in more risk-adverse health behaviors.

Three health risk behaviors are identified: body mass index (BMI), current smoking, and current alcohol consumption. The NHWS does not allow a more detailed assessment of actual alcohol consumption or number of cigarettes per day and duration of smoking behavior.

In China, we would expect the presence of rheumatoid arthritis to be associated negatively with workforce status as well as absenteeism and presenteeism. At the same time, health risk behaviors, at least in respect of the presence of obesity and morbid obesity, would be expected to have a negative impact. The Charlson co-morbidity index would also be expected to have a negative impact27. A key question is whether the presence of rheumatoid arthritis has a quantitatively greater impact than these other health status attributes.

Because of the ambiguous role of family income as both an outcome and a potential determinant of labor force activity, as well as absenteeism and presenteeism, this variable is dropped from the estimated models.

Estimates of the independent variables are presented in Table 7. Note that the rheumatoid arthritis population is older than controls (notably the percentage in the 40–59 years age group); a higher proportion of the rheumatoid arthritis sample are females; household income is somewhat greater for the rheumatoid arthritis sample; there is a tendency towards a higher BMI in the rheumatoid arthritis sample with a substantially higher percentage reporting smoking behavior. Most noteworthy, perhaps, is the substantially higher CCI in the rheumatoid arthritis sample (1.448 vs 0.244).

Table 6.  Workplace productivity and activity impairment scale: Impact of health problems on presenteeism in the past 7 days, urban China, NHWS 2009.

Presenteeism responses	Rheumatoid Arthritis (%)	Control population (%)
0 (Health problems had no effect on my work)	26 (11.61)	2715 (30.73)
1	22 (9.82)	1783 (20.18)
2	39 (17.41)	1358 (15.37)
3	35 (15.63)	986 (11.16)
4	21 (9.38)	412 (4.66)
5	23 (10.27)	562 (6.36)
6	32 (14.29)	485 (5.49)
7	18 (8.04)	309 (3.50)
8	5 (2.23)	126 (1.43)
9	3 (1.34)	56 (0.63)
10 (Health problems completely prevented me from working)	0 (0.00)	42 (0.48)

Source: NHWS, 2009.

Table 7.  Non-pain independent variables for regression modeling, urban China, NHWS 2009.

Independent variables	Rheumatoid Arthritis (%)	Control population (%)
Socio-economic characteristic
Age: 18–39	116 (32.86)	6324 (48.82)
Age: 40–59	181 (51.27)	4680 (36.13)
Age: 60+	56 (15.86)	1950 (15.05)
Male	163 (46.18)	6757 (52.16)
Education: Other	74 (20.96)	2149 (16.59)
Education: High School	112 (31.73)	4658 (35.96)
Education: University	167 (47.31)	6147 (47.45)
Household Income:
Under RMB 3000	122 (34.56)	5104 (39.40)
RMB 4000–7999	126 (35.69)	3969 (30.64)
RMB 8000+	100 (28.33)	3343 (25.81)
Declined to answer	5 (1.42)	538 (4.15)
Health risk behaviors
BMI: Normal	198 (56.09)	7979 (61.59)
BMI: Under weight	30 (8.50)	1174 (9.06)
BMI: Overweight	91 (25.78)	2636 (20.35)
BMI: Obese	13 (3.68)	428 (3.30)
BMI: Morbidly Obese	19 (5.38)	523 (4.04)
BMI: Not estimated	2 (0.57)	214 (1.65)
Alcohol use	184 (52.12)	5026 (38.80)
Smoking	100 (28.33)	3122 (24.10)
Charlson comorbidity index (SD)	1.448 (0.793)	0244 (0.711)

Source: NHWS 2009.

Modeling considerations

Two issues are important in assessing the impact of rheumatoid arthritis on HRQoL and health status. These are: (i) the use of sampling weights in the regression models and (ii) the choice of regression model.

Sampling weights in regression models

In the case where sampling weights are a function of independent variables (in this NHWS age and gender) the traditional view is that unweighted regression models are to be preferred28. On the assumption that the model is correctly specified, both unweighted and weighted regressions will generate unbiased and consistent parameter estimates. To the extent that parameter estimates differ between weighted and unweighted models this may be indicative of model misspecification. In this report the regression models are unweighted. However, as a check on the results, regressions were also estimated using probability weights. There was no difference in results. STATA v. 11 was used for all regressions.

Choice of regression model

The choice of regression estimator (generalized linear model) is determined by the characteristics of the dependent variable. In this analysis, as the HRQoL variables are continuous with no floor or ceiling effects, ordinary least squares (OLS) is employed. Given that health status is categorical and ordered, an ordered logit model is used. In the case of labor force status, as the dependent variable is binary, the model of choice is logistic while a multinomial logit model is used to predict the impact of rheumatoid arthritis on labor force status. Finally, in the absenteeism and presenteeism models, as the dependent variables are ranked (from least affected to most affected) an ordered logit model is employed.

Results

Health-related quality-of-life

Regression results for PCS model are given in Table 8. The presence of rheumatoid arthritis is significant at the 5% level and enters with the expected negative sign (−2.289; 95%CI: −3.042 to −1.536). The impact of rheumatoid arthritis is greater than that for the presence of co-morbidities using the CCI (−1.882; 95%CI: −2.046 to −1.717). Health risk behaviors are all significant at the 5% level and enter with negative signs. The most important influence is for persons 60 years and over (−5.905; 95%CI: −6.311 to −5.499).

Table 8.  SF-12 physical component summary scores: Regression results, urban China, NHWS 2009.

Independent variables	Coefficient	‘t’ statistic	95% Confidence Interval
Presence of rheumatoid arthritis	−2.289	−5.96	−3.042	−1.536
Socio-demographic variables
Age: 40–59 years^a	−1.717	−12.29	−1.991	−1.443
Age: 60 years+^a	−5.905	−28.50	−6.311	−5.499
Gender: male^b	1.805	12.89	1.530	2.079
Education: University or higher^c	0.973	4.94	0.586	1.359
Education: High school completed^c	1.420	6.72	1.006	1.834
Income: R20,000–R39,999^c	0.064	0.43*	−0.227	0.354
Income: R40,000+^d	−0.284	−1.73*	−0.605	0.037
Health Risk Behaviors
BMI: Underweight^e	−0.986	−4.59	−1.407	−0.565
BMI: Overweight^e	−0.676	−4.38	−0.979	−0.374
BMI: Obese^e	−1.717	−5.12	−2.374	−1.059
BMI: Morbidly obese^e	−0.965	−3.16	−1.563	−0.367
Current smoker^f	−0.305	−1.94*	−0.614	0.003
Alcohol user^g	−0.289	−2.11	−0.557	−0.021
Morbidity/Comorbidity status
Charlson Comorbidity Index	−1.882	−22.43	−2.046	−1.717
Constant	49.270	225.21	48.841	49.698
Adj R^2	0.160
N	13,307

*Not significant at 5% level.

Reference categories: ^aage 18–39 years; ^bfemales; ^cnot completed high school; ^dincome under R20,000; ^eBMI normal weight; ^fnon-smoker; ^gnon-drinker.

Regression results for MCS are given in Table 9. The impact of the presence of rheumatoid arthritis on MCS is less than that for PCS (−1.472; 95%CI: −2.338 to −0.605). Unlike the PCS model, age enters with a positive sign. The CCI enters with the expected negative sign (−1.471; 95%CI: −1.660 to −1.281). Results for health risk behaviors are mixed. Being underweight and morbidly obese enter with the expected negative signs, as does the use of alcohol.

Table 9.  SF-12 mental component summary scores: Regression results, urban China, NHWS 2009.

Independent variables	Coefficient	‘t’ statistic	95% Confidence Interval
Presence of rheumatoid arthritis	−1.472	−3.33	−2.338	−0.605
Socio-demographic variables
Age: 40–59 years^a	3.008	18.70	2.692	3.323
Age: 60 years+^a	3.831	16.06	3.364	4.299
Gender: male^b	1.240	7.69	0.924	1.556
Education: University or higher^c	−1.030	−4.54	−1.474	−0.585
Education: High school completed^c	−1.581	−6.50	−2.058	−1.105
Income: R20,000–R39,999^d	−0.062	−0.37*	−0.397	0.272
Income: R40,000+^d	0.022	0.12*	−0.348	0.391
Health Risk Behaviors
BMI: Underweight^e	−1.378	−5.57	−1.863	−0.893
BMI: Overweight^e	0.414	2.33	0.066	0.763
BMI: Obese^e	−0.140	−0.36*	−0.897	0.617
BMI: Morbidly obese^e	−0.852	−2.43	−1.540	−0.163
Current smoker^f	−0.191	–1.05*	−0.546	0.164
Alcohol user^g	−1.249	−7.94	−1.557	−0.940
Morbidity/Comorbidity Status
Charlson Comorbidity Index	−1.471	−15.22	−1.660	−1.281
Constant	45.746	181.61	45.252	46.240
Adj R^2	0.092
N	13,307

*Not significant at 5% level.

Reference categories: ^aage 18–39 years; ^bfemales; ^cnot completed high school; ^dincome under R20,000; ^eBMI normal weight; ^fnon-smoker; ^gnon-drinker.

Regression results for the absolute utility scores are given in Table 10. The presence of rheumatoid arthritis has a negative and statically significant impact on SF-6D utilities (−0.025; 95%CI: −0.036 to −0.014). This impact is substantially greater than for health risk behaviors. The negative impact is somewhat greater than for the CCI (−0.021; 95%CI: −0.023 to −0.0.019).

Table 10.  SF-6D absolute utilities: Regression results, urban China, NHWS 2009.

Independent variables	Coefficient	‘t ’ statistic	95% Confidence Interval
Presence of rheumatoid arthritis	−0.025	−4.59	−0.036	−0.014
Socio-Demographic Variables
Age: 40–59 years^a	0.026	12.85	0.022	0.030
Age: 60 years+^a	0.014	4.70	0.008	0.020
Gender: male^b	0.020	9.76	0.016	0.023
Education: University or higher^c	−0.004	−1.37*	−0.009	0.002
Education: High school completed^c	−0.010	−3.46	−0.016	−0.004
Income: R20,000–R39,999^d	0.004	2.09	0.0002	0.009
Income: R40,000+^d	0.002	0.97*	−0.002	0.007
Health Risk Behaviors
BMI: Underweight^e	−0.015	−4.89	−0.021	−0.009
BMI: Overweight^e	0.005	2.47	0.001	0.010
BMI: Obese^e	−0.006	−1.16*	−0.015	0.004
BMI: Morbidly obese^e	−0.011	−2.42	−0.019	−0.002
Current smoker^f	0.002	0.69*	−0.003	0.006
Alcohol user^g	−0.016	−8.33	−0.020	−0.012
Morbidity/Comorbidity Status
Charlson Comorbidity Index	−0.021	−17.47	−0.023	−0.019
Constant	0.696	221.45	0.690	0.702
Adj R^2	0.098
n	13,307

*Not significant at 5% level.

Reference categories: ^aage 18–39 years; ^bfemales; ^cnot completed high school; ^dincome under R20,000; ^eBMI normal weight; ^fnon-smoker; ^gnon-drinker.

Health status

Regression results for the health status responses are given in Table 11. A negative impact on health status (given the ordering of the health status responses) is indicated by an odds ratio greater than one. The impact of rheumatoid arthritis is significant (odds ratio 1.275; 95%CI: 1.031 to 1.576). The CCI has a significantly greater impact (odds ratio 1.819; 95%CI: 1.711 to 1.933). Health risk factors have a mixed impact with being underweight, overweight, and morbidly obese having a significant negative impact as well as alcohol consumption.

Table 11.  Health status: Regression results, urban China, NHWS 2009.

Independent variables	Odds ratio	p > │z│	95% Confidence Interval
Presence of rheumatoid arthritis	1.275	0.03	1.031	1.576
Socio-Demographic Variables
Age: 40–59 years^a	1.286	0.00	1.195	1.384
Age: 60 years+^a	2.502	0.00	2.235	2.801
Gender: male^b	0.641	0.00	0.595	0.690
Education: University or higher^c	0.836	0.00	0.752	0.929
Education: High school completed^c	0.678	0.00	0.605	0.758
Income: R20,000–R39,999^d	0.997	0.93*	0.922	1.078
Income: R40,000+^d	0.928	0.09*	0.851	1.011
Health Risk Behaviors
BMI: Underweight^e	1.198	0.00	1.069	1.342
BMI: Overweight^e	1.266	0.00	1.167	1.374
BMI: Obese^e	1.379	0.00	1.151	1.652
BMI: Morbidly obese^e	0.994	0.94*	0.845	1.170
Current smoker^f	1.099	0.03	1.011	1.194
Alcohol user^g	0.859	0.00	0.800	0.923
Morbidity/Comorbidity Status
Charlson Comorbidity Index	1.819	0.00	1.711	1.933
Loglikelihood	−16,995.13
Pseudo R^2	0.042
n	13,307

*Not significant at 5% level.

Reference categories: ^aage 18–39 years; ^bfemales; ^cnot completed high school; ^dincome under R20,000; ^eBMI normal weight; ^fnon-smoker; ^gnon-drinker.

Labor force status

The logit model results for labor force status are presented in Table 12. The presence of rheumatoid arthritis reduces the probability of labor force participation (odds ratio 0.675; 95%CI: 0.488 to 0.932). Male gender and educational attainment, as expected, increases the probability of participation. Among the health risk factors morbid obesity, alcohol use, and smoking have a positive association, as does the presence of high BMIs.

Table 12.  Labor force participation: Regression results, urban China, NHWS, 2009.

Independent variables	Odds ratio	p > |z|	95% Confidence Interval
Presence of rheumatoid arthritis	0.675	0.02	0.488	0.932
Socio-Demographic Variables
Age: 40–59 years^a	0.360	0.00	0.319	0.407
Age: 60 years+^a	0.012	0.00	0.010	0.015
Gender: male^b	1.616	0.00	1.426	1.832
Education: University or higher^c	4.444	0.00	3.829	5.158
Education: High school completed^c	8.817	0.00	7.522	10.334
Health Risk Behaviors
BMI: Underweight^d	0.881	0.19*	0.730	1.064
BMI: Overweight^d	0.885	0.07*	0.774	1.012
BMI: Obese^d	1.105	0.52*	0.815	1.496
BMI: Morbidly obese^d	1.623	0.00	1.195	2.203
Current smoker^e	1.410	0.00	1.211	1.641
Alcohol user^f	2.187	0.00	1.928	2.481
Morbidity/Comorbidity Status
Charlson Comorbidity Index	1.078	0.10*	0.985	1.179
Loglikelihood	−4638.87
Pseudo R^2	0.420
n	13,307

*Not significant at 5% level.

Reference categories: ^aage 18–39 years; ^bfemales; ^cnot completed high school; ^dBMI normal weight; ^enon-smoker; ^fnon-drinker.

Predicted labor force status

Utilizing a multinomial logit approach it is possible to estimate the relative probabilities of the impact of rheumatoid arthritis on labor force participation categories. These results are presented in Table 13. The results show that the presence of rheumatoid arthritis has a pronounced impact on labor force status. Persons with rheumatoid arthritis report an estimated 68.09% probability of labor force participation rate compared to 75.92% for those in the control population. The probability of being employed full-time is less (61.30% vs 64.40%), with similar deficits for being employed part-time and self-employed.

Table 13.  Labor force participation regression results: Predicted relative probabilities by rheumatoid arthritis status, urban China, NHWS 2009.

Labor force status	Rheumatoid Arthritis (%)	Control population (%)
Employed full-time	61.30	64.60
Employed part-time	2.52	4.89
Self-employed	2.70	4.78
Unemployed	1.57	1.66
In the labor force	68.09	75.92
Not in the labor force	31.91	24.08
Total	100.00	100.00

Absenteeism

The presence of rheumatoid arthritis has a significant negative impact on reported absenteeism (Table 14). With an odds ratio of 1.547 (95%CI: 1.167 to 2.050), it is only alcohol use that has a similar impact. All health risk factors have a negative impact and are statistically significant with the exception of being overweight. This holds true for the Charlson Co-morbidity Index. Younger age, male gender, and higher educational attainment reduce absenteeism. The variable with the greatest impact on absenteeism is alcohol use (odds ratio: 1.692: 95%CI 1.500 to 1.908).

Table 14.  Absenteeism: Regression results, urban China, NHWS 2009.

Independent variables	Odds ratio	p > |z|	95% Confidence Interval
Presence of rheumatoid arthritis	1.547	0.00	1.167	2.050
Socio-Demographic Variables
Age: 40–59 years^a	0.686	0.00	0.607	0.774
Age: 60 years+^a	1.047	0.84*	0.675	1.625
Gender: male^b	0.613	0.00	0.538	0.698
Education: University or higher^c	0.689	0.01	0.524	0.904
Education: High school completed^c	0.853	0.24*	0.653	1.114
Health Risk Behaviors
BMI: Underweight^d	1.226	0.03	1.022	1.471
BMI: Overweight^d	1.018	0.82*	0.875	1.183
BMI: Obese^d	1.202	0.23*	0.893	1.618
BMI: Morbidly obese^d	1.540	0.00	1.223	1.939
Current smoker^e	1.356	0.00	1.183	1.554
Alcohol user^f	1.692	0.00	1.500	1.908
Morbidity/Comorbidity Status
Charlson Comorbidity Index	1.410	0.00	1.321	1.505
Loglikelihood	−6568.23
Pseudo R^2	0.0293
n	8980

*Not significant at 5% level.

Reference categories: ^aage 18–39 years; ^bfemales; ^cnot completed high school; ^dBMI normal weight; ^enon-smoker; ^fnon-drinker.

Presenteeism

The presence of rheumatoid arthritis increases the incidence of presenteeism (odds ratio 1.588; 95%CI: 1.250 to 2.017) (Table 15). This is similar to the impact of chronic comorbidities as measured by the Charlson score (odds ratio: 1.535; 95%CI: 1.445 to 1.631). Alcohol use also has a major impact on presenteeism (odds ratio 1.747; 95% CI: 1.610 to 1.895). Completing high school also has a substantive impact compared to the reference group, and greater than for those reporting post-high school education.

Table 15.  Presenteeism: Regression results, urban China, NHWS 2009.

Independent variables	Odds ratio	p > |z|	95% Confidence Interval
Presence of rheumatoid arthritis	1.588	0.00	1.250	2.017
Socio-Demographic Variables
Age: 40–59 years^a	0.793	0.00	0.732	0.859
Age: 60 years+^a	1.051	0.76*	0.766	1.444
Gender: male^b	0.622	0.00	0.570	0.678
Education: University or higher^c	1.119	0.24*	0.928	1.348
Education: High school completed^c	1.469	0.00	1.221	1.766
Health Risk Behaviors
BMI: Underweight^d	1.398	0.00	1.228	1.593
BMI: Overweight^d	0.944	0.26*	0.854	1.043
BMI: Obese^d	1.116	0.31	0.904	1.379
BMI: Morbidly obese^d	1.256	0.10	1.055	1.495
Current smoker^e	1.050	0.31*	0.955	1.155
Alcohol user^f	1.747	0.00	1.610	1.895
Morbidity/Comorbidity Status
Charlson Comorbidity Index	1.535	0.00	1.445	1.631
Loglikelihood	−17,208.50
Pseudo R^2	0.022
n	9085

*Not significant at 5% level.

Reference categories: ^aage 18–39 years; ^bfemales; ^cnot completed high school; ^dBMI normal weight; ^enon-smoker; ^fnon-drinker.

Discussion

The results for HRQoL and self-reported health status are robust. In all cases the presence of rheumatoid arthritis is significant and is associated with lower quality-of-life, self-reported health status, and employment. Quantitatively, it is one of the most important variables in all the models. In the case of the PCS model, it is second only to the impact of age in its impact on the PCS score. In the MCS model it is also significant but of a similar quantitative impact to the CCI. In the SF-6D utility model, the presence of rheumatoid arthritis is the most important variable in its deficit effect on utility scores, followed by the CCI. In the case of self-reported health status, the impact of rheumatoid arthritis, while still significant and quantitatively important, is eclipsed by the CCI and the effect of age.

At the same time the analysis supports the hypothesis that the presence of rheumatoid arthritis reduces both the probability of labor force participation as well as increasing likely absenteeism and presenteeism in the urban Chinese populations. These results, which appear to be robust, point to the key role of the presence of rheumatoid arthritis compared to the presence of health risk behaviors – notably alcohol use. In both absenteeism and presenteeism models, the impact of rheumatoid arthritis is similar to that for the Charlson score and, at least for the absenteeism model, similar to the impact of health risk behaviors.

As this is the first time national estimates of the impact of rheumatoid arthritis have been generated for urban China, it is impossible to come to any conclusions as to the impact of rheumatoid arthritis vis-à-vis other disease states. Even so, with the control population as a reference category, it is clear that the presence of rheumatoid arthritis has a significant deficit impact on the aspects of disease burden considered. Given how little we know about the correlates of the burden of disease in urban China, these results should be seen as only a first attempt to evaluate the deficit impact of a prevalent disease state. There is, clearly, a substantial research agenda to be considered in respect, not only of the burden of illness associated with other highly prevalent disease states, but also the potential independent effect of comorbidities (e.g., pain) that are associated with the underlying inflammatory conditions characterizing rheumatoid arthritis.

Limitations of the analysis

As the NHWS is an internet-based survey, biases may be present due to the lack of internet penetration in the urban China population. The extent to which individuals and households have internet access is unknown. In addition, the NHWS relies upon respondents reporting they have been diagnosed with one or more specific disease states. These are not, given the nature of the survey, clinically verified. This also introduces a degree of uncertainty. In the absence of other national health surveys in China, surveys such as the NHWS fill an important gap. As the NHWS covers some 140 disease states, there is clearly a lot more that can be done to assess the contribution of other chronic disease states to HRQoL and self-reported health status in this country. Even so, given the prevalence estimate reported here, care should be taken in uncritically generalizing these results to the wider China population.

Conclusions

The results presented here represent for the first time a comprehensive, population-based assessment of the prevalence, characteristics, and impact of self-reported rheumatoid arthritis on HRQoL, health status, employment, and productivity for urban China. The results suggest that the presence of rheumatoid arthritis, including attendant co-morbidities, imposes a substantial disease burden. A burden which is most clearly seen in terms of HRQoL and health status deficits.

Declaration of interest

Paul Langley is a consultant to Kantar Health, who were paid consultants to Pfizer in connection with the development of this manuscript; Rong Mu is a consultant to Pfizer, China; Michael Wu and Peng Dong are employees of Pfizer, China; Boxiong Tang is an employee of Pfizer Inc, USA.

Transparency

Declaration of funding

This study was supported by Pfizer Inc, USA.

Acknowledgments

No assistance in the preparation of this article is to be declared.