Burden of musculoskeletal disease and its determination by urbanicity, socioeconomic status, age, and sex: Results from 14,507 subjects

Authors


Abstract

Objective

The availability of reliable estimates of the burden of musculoskeletal disease is of considerable importance for policymakers.

Methods

This study uses data from the 14,507 participants of the European Health Interview Survey conducted in Austria in 2006/2007 to calculate estimates of the prevalence of osteoarthritis, spinal conditions, and osteoporosis in a population representative of other European Union or Organisation for Economic Co-operation and Development member states. Urbanicity, socioeconomic status, and age and sex were included as determinants of musculoskeletal disease.

Results

The prevalence of arthritis was 18.8% (95% confidence interval [95% CI] 18.2–19.4%), of spinal conditions was 38.4% (95% CI 37.6–39.2%), and of osteoporosis was 6.6% (95% CI 6.3–7.0%). The census data showed strong evidence for an association between urbanicity and arthritis (P = 0.012) and osteoporosis (P < 0.001), but not spinal conditions (P = 0.721). Arthritis and spinal conditions were associated with socioeconomic status (P < 0.001 for all). Osteoporosis showed the same associations with age, income, and education. For arthritis, a combined model showed a substantial attenuation of the effect of urbanicity on arthritis prevalence after adjustment for socioeconomic status.

Conclusion

These data suggest that the burden of musculoskeletal disease is determined by both urbanicity and socioeconomic status; however, the effect of urbanicity seems to be attributable to differences in socioeconomic status and demographics across geographic regions.

INTRODUCTION

Musculoskeletal disease is among the most common causes of severe long-term disability and protracted pain in industrialized societies (1, 2). It is commonly accepted that musculoskeletal disease has a particularly high prevalence and thus a large impact on industrialized nations and advanced economies (3). The impact and importance of musculoskeletal diseases are critical not only for individual health and mobility, but also for social functioning and productivity and economic growth on a larger scale (4, 5). This importance has been acknowledged internationally by officials and policymakers as illustrated by the proclamation of the Bone and Joint Decade 2000–2010 (6, 7).

Given this importance of musculoskeletal disorders on both the individual and population levels, it is crucial to have reliable estimates of the burden of musculoskeletal disorders. However, current estimates of the burden of musculoskeletal disease are either highly focused research articles (i.e., one disease in one specific group such as hospital workers [8, 9]) or extensive reports on the global situation such as the World Health Organization (WHO) Technical Report on musculoskeletal conditions (online at http://whqlibdoc.who.int/trs/WHO_TRS_919.pdf). However, it seems there is somewhat of a lack of data in the middle ground, i.e., large but nevertheless homogenous populations, especially so for industrialized countries.

Beyond describing the burden of musculoskeletal disease per se, it is equally important to recognize its major determinants in order to be able to predict and maybe even influence trends. Environmental and behavioral factors such as socioeconomic status, sedentary lifestyle, and type of profession are very likely to play a role in the burden of musculoskeletal disease. Living in an urban environment, referred to as urbanicity, has been suggested to be another major factor in the development of disease (10–13). Again, while there are a number of publications on this issue for developing countries (14–17), there seems to be less focus on the systematic description of these determinants in large populations representative of industrialized countries.

The first objective of this study was to describe the current burden of musculoskeletal disease in Austria as a representation of advanced economies and European Union (EU) or Organisation for Economic Co-operation and Development (OECD) member states. The burden of these diseases will be shown by using the prevalence of arthritis, spinal conditions, and osteoporosis, and also by measuring medication use as a result of these diseases. Second, the objective of this study was to investigate the effect of urbanicity, socioeconomic status, and age and sex on the prevalence of musculoskeletal conditions.

Significance & Innovations

  • Seven percent to 40% of all Austrians have musculoskeletal disease, and approximately one-half of them require medication.

  • Musculoskeletal disease is not associated with urbanicity, but with socioeconomic status and demographic variables.

  • Addressing socioeconomic status and demographic variables could reduce the burden of musculoskeletal disease by up to 10%.

MATERIALS AND METHODS

Survey.

In 2004, Eurostat, the statistical office of the EU, and the Directorate General for Health and Consumer Protection of the EU commissioned the European Health Survey System (EHSS), which combines regularly conducted surveys as well as special surveys, database creation, and methodologic issues. Once fully implemented it is anticipated that the EHSS will collect data regularly in a systematic and comprehensive fashion for the approximately 500 million inhabitants of the 27 EU member states. The European Health Interview Survey (EHIS) serves as a vehicle survey that will be used across the EU every 5 years with additional questions at each member state's discretion. The first use of the EHIS was in 2006/2007 in Austria (ATHIS) and Estonia.

For the ATHIS 2006/2007, a study frame was defined as all registered inhabitants of Austria ages ≥15 years, regardless of nationality (6.9 million). The Austrian health care system defines 32 administrative regions based on political and geographic units. A sample of the total population was chosen in such a way as to keep the variance across the 32 regions equal, i.e., to achieve equal estimate precision across the 32 regions of different sizes. The survey was conducted over 1 year to compensate for seasonal difference in responses (e.g., subjective health during flu season). Computer-aided personal interviews were conducted by specifically trained interviewers. This allowed for double entry with immediate cross-checking for all variables. A second plausibility test was done after all data were collected. Unit nonresponse was dealt with by weighting class adjustments, using data on population of the area of residence of the unit nonresponse, as well as age groups, sex, and nationality. Item nonresponse (less than 8%) was dealt with by hot-deck imputation of missing values, whereby the missing item is transferred from a different participant of the same age, sex, level of education, and area of residence. The results of the ATHIS are published in aggregate form on the web site of the Austrian statistical office Statistic Austria (online at www.statistik.at). Individuals and academic institutions can apply for access to anonymized microlevel data for scientific analysis.

Included variables.

Data on musculoskeletal disease were extracted for 14,507 participants of the ATHIS. Included in this study are prevalences of arthritis, spinal conditions, and osteoporosis. Participants were offered a list of chronic diseases and asked to identify their chronic and actively ongoing problems of the last 12 months. Arthritis, spinal conditions, and osteoporosis were the 3 musculoskeletal diseases included in this list. To demonstrate the impact of these diseases on the Austrian health care system, the prevalence of medication use for these disease entities was included as well. Medication use was defined as using medication (prescribed or over the counter) for the abovementioned conditions within the 2 weeks prior to the survey.

The following potential determinants of the burden of musculoskeletal disease were included: the Nomenclature of Territorial Units for Statistics III classification of urbanicity (18), ranging from 1 (rural area) to 3 (urban area), socioeconomic status reflected by income level (15 levels), and highest level of education. The 5 levels of education are given as a range from “lowest” to “highest” since the technical terms for the individual steps of the Austrian educational system cannot readily be translated into English. Lowest education is equivalent to legally mandatory schooling until age 15 years, and highest education is equivalent to an advanced degree. Additionally, demographic data on age (in groups with a width of 5 years) and sex were included as potential confounders and to test for potential sex inequality.

Statistical modeling.

All data extractions were double checked for errors. All prevalences are given with binominal 95% confidence intervals (95% CIs) for the total studied sample and aggregated by sex. The effect of sex on disease prevalences is given as odds ratio (OR; 95% CI).

To test for the influence of urbanicity, socioeconomic status, and age and sex, multivariate generalized logit regression models are used to calculate ORs. Income, age group, sex, and highest level of education are tested for their role as predictors as well as potential confounders and/or interacters in the association between musculoskeletal disease and urbanicity.

An alpha level of 5% was considered significant. However, given the substantial size of the study population, standard errors are expected to be very small, leading to very low P values as an effect of sample size. Therefore, this study emphasizes CIs to better describe the size and thus clinical and social impact of effects rather than their statistical significance. All calculations were done using Stata, version 10 (StataCorp).

RESULTS

Study sample.

A total of 14,507 individuals, 7,938 women and 6,569 men, were included in this survey. The mean age was 36 years (95% CI 35.7–36.3 years). By area, 2,183 (15.05%) live in an urban area, 7,951 (54.81%) live in a rural area, and 4,373 (30.14%) live an intermediate area.

Disease prevalence and medication use.

Arthritis.

The overall prevalence of arthritis in the full sample was 18.8% (95% CI 18.2–19.4%), or 15.1% (95% CI 14.2–16.1%) for men and 21.8% (95% CI 20.9–22.8%) for women (Figure 1). This sex difference is consistent with an OR of 0.63 (95% CI 0.58–0.69), which is statistically significant (P < 0.001). Among the 18.8% or 2,727 individuals with arthritis, the prevalence of medication users was 59.1% (95% CI 57.3–61.0%), or 61.1% (95% CI 58.8–63.4%) for women and 55.8% (95% CI 52.7–58.9%) for men, resulting in an OR of 0.80 (95% CI 0.69–0.94, P = 0.007).

Figure 1.

Prevalence of arthritis by sex.

Spinal conditions.

Spinal conditions had an overall prevalence of 38.4% (95% CI 37.6–39.2%) in the studied population. Aggregated by sex, there was a prevalence of 39.3% (95% CI 38.2–40.4%) for women and 37.2% (95% CI 36.1–38.4%) for men, with an OR of 0.92 (95% CI 0.86–0.98), i.e., significantly higher odds of spine problems in women (P = 0.010) (Figure 2). Of individuals with spinal conditions, 45.5% (95% CI 44.2–46.9%) used medication for these problems. As before, medication use was higher in women, with 49.6% (95% CI 47.8–51.3%), than in men, with 40.3% (95% CI 38.3–42.2%). This sex difference is consistent with an OR of 0.69 (95% CI 0.62–0.76) and is therefore statistically significant (P < 0.001).

Figure 2.

Prevalence of spinal conditions by sex.

Osteoporosis.

Of the studied population, 6.6% (95% CI 6.3–7.0%) had osteoporosis. Again, women predominated, with a prevalence of 10.6% (95% CI 9.9–11.3%), over men, with a prevalence of 1.9% (95% CI 1.6–2.3%) (Figure 3). This obvious sex difference is consistent with an OR of 0.17 (95% CI 0.14–0.20), which is statistically significant (P < 0.001). Of all individuals with osteoporosis, 80.2% (95% CI 77.6–82.7%) received medication for this condition. The prevalence of medication users in women was 81.1% (95% CI 78.4–83.0%) and in men was 74.2% (95% CI 66.5–81.9%), resulting in an OR of 0.67 (95% CI 0.44–1.04), which is not statistically significant (P = 0.071).

Figure 3.

Prevalence of osteoporosis by sex.

Determinants of the burden of disease.

Urbanicity.

The univariate assessment of the census data showed strong evidence for an association between urbanicity and arthritis (P = 0.012) as well as osteoporosis (P < 0.001), consistent with a decrease in disease prevalence with increasing urbanicity. There was no evidence for an association between urbanicity and spinal conditions (P = 0.721). There was no evidence for an effect of urbanicity on medication use for arthritis (P = 0.313) or osteoporosis (P = 0.498). There was a significant (P = 0.043) association between urbanicity and medication usage for spinal conditions, which was lower in areas with higher levels of urbanicity.

Socioeconomic status.

A regression model including both income and education showed that arthritis was associated with income and highest level of education attained, with a P value of <0.001 for both, while showing a decrease in prevalence with higher socioeconomic status (Table 1). The same was seen for medication use for arthritis (Table 2). Spinal conditions were associated with education but not income, while medication use for spinal conditions was associated with both (Table 2). Osteoporosis decreased with income and education level, yet medication use for osteoporosis was associated only with education (Table 2).

Table 1. Odds of disease by urbanicity, SES, demographics, and all combined*
 UrbanicitySESDemographicsCombined
OR (95% CI)POR (95% CI)POR (95% CI)POR (95% CI)P
  • *

    SES = socioeconomic status; OR = odds ratio; 95% CI = 95% confidence interval.

Arthritis        
 Urban-rural area0.93 (0.88–0.98)0.0121.02 (0.95–1.09)0.532
 Income group0.96 (0.94–0.98)< 0.0010.96 (0.95–0.98)< 0.001
 Education0.84 (0.81–0.88)< 0.0010.84 (0.81–0.88)< 0.001
 Age group1.33 (1.31–1.35)< 0.0011.33 (1.31–1.35)< 0.001
 Sex0.74 (0.68–0.82)< 0.0010.74 (0.66–0.81)< 0.001
Spinal conditions        
 Urban-rural area0.99 (0.95–1.04)0.7211.01 (0.96–1.06)0.712
 Income group1.01 (1.00–1.02)0.1981.01 (1.00–1.02)0.200
 Education0.94 (0.92–0.97)< 0.0010.94 (0.92–0.97)< 0.001
 Age group1.15 (1.14–1.16)< 0.0011.15 (1.14–1.16)< 0.001
 Sex0.99 (0.92–1.06)0.6970.98 (0.91–1.06)0.629
Osteoporosis        
 Urban-rural area0.73 (0.66–0.81)< 0.0011.05 (0.92–1.19)0.483
 Income group0.97 (0.94–1.00)0.0550.97 (0.94–1.00)0.053
 Education0.92 (0.86–0.97)0.0200.92 (0.86–0.99)0.020
 Age group1.39 (1.36–1.43)< 0.0011.39 (1.36–1.43)< 0.001
 Sex0.19 (0.15–0.23)< 0.0010.18 (0.15–0.22)< 0.001
Table 2. Odds of medication use by urbanicity, SES, demographics, and all combined*
 UrbanicitySESDemographicsCombined
OR (95% CI)POR (95% CI)POR (95% CI)POR (95% CI)P
  • *

    SES = socioeconomic status; OR = odds ratio; 95% CI = 95% confidence interval.

Arthritis        
 Urban-rural area0.94 (0.85–1.05)0.3130.99 (0.88–1.11)0.827
 Income group0.97 (0.93–1.00)0.0390.97 (0.93–1.00)0.039
 Education0.90 (0.83–0.97)0.0060.89 (0.83–0.97)0.006
 Age group1.05 (1.02–1.08)< 0.0011.05 (1.02–1.08)< 0.001
 Sex0.89 (0.75–1.04)0.1470.89 (0.75–1.06)0.194
Spinal conditions        
 Urban-rural area0.93 (0.86–0.998)0.0431.02 (0.94–1.10)0.650
 Income group0.98 (0.95–1.00)0.0200.98 (0.94–0.81)0.020
 Education0.87 (0.83–0.92)< 0.0010.87 (0.83–0.92)< 0.001
 Age group1.05 (1.03–1.07)< 0.0011.05 (1.03–1.07)< 0.001
 Sex0.72 (0.65–0.81)< 0.0010.72 (0.65–0.81)< 0.001
Osteoporosis        
 Urban-rural area1.1 (0.83–1.45)0.4981.2 (0.91–1.6)0.192
 Income group0.98 (0.91–1.05)0.5640.98 (0.91–1.05)0.517
 Education0.85 (0.72–0.996)0.0440.85 (0.73–0.996)0.045
 Age group1.06 (0.98–1.13)0.1401.06 (0.98–1.14)0.124
 Sex0.82 (0.52–1.29)0.3830.76 (0.48–1.21)0.253

Age and sex.

In a multivariate regression including age and sex, arthritis was associated with both age and sex (Table 1). Medication for arthritis use was associated with age but not with sex (Table 2). Spinal conditions were associated with higher age, but not with sex. The use of medication for spinal conditions was associated with both age and sex, again with higher use in older and female participants (Table 2). Osteoporosis prevalence increased with age and was lower in men than in woman. Medication use for osteoporosis was not associated with age or sex (Table 2).

Combined effects of all determinants.

A multivariate regression model assessed the combined effects of urbanicity, socioeconomic status, age, and sex (Table 1).

For arthritis, the combined model showed a substantial attenuation of the effect of urbanicity on arthritis prevalence to statistically nonsignificant values. The effect of socioeconomic status remained unchanged after inclusion of urbanicity into the regression model, suggesting that these parameters act independently from urbanicity.

The effect of urbanicity on spinal conditions was not affected by socioeconomic status. At the same time, the effects of socioeconomic status were not affected by inclusion of a measure of urbanicity.

As before, the trend for osteoporosis was very similar to the results for arthritis. Inclusion of socioeconomic status reduced the coefficient for urbanicity on osteoporosis into the range of nonsignificant values, providing no evidence for an effect of urbanicity on the prevalence of osteoporosis after accounting for socioeconomic status. In return, there was no evidence that the effect of socioeconomic status on osteoporosis prevalence is confounded by urbanicity, other than a minimal reduction in the effect of sex.

DISCUSSION

Access to reliable estimates of the current burden of musculoskeletal disease and its determinants is highly important for policymakers to plan and direct trends in public health and health care finance. This study aimed at providing such estimates for a current large population that can be regarded as a valid representation of industrialized nations, first by describing the prevalence of musculoskeletal diseases and measuring medication use as a result of these diseases, and second by investigating the effect of socioeconomic status, urbanicity, and age and sex on disease prevalence and medication use. Briefly, we found that between 7% and 40% of the population of this study was affected by at least 1 of the 3 included diseases: arthritis, spinal conditions, or osteoporosis. Fifty percent to 80% of these patients use medication to manage these problems. An initially seen association between disease prevalence and urbanicity proved to be confounded by socioeconomic and demographic differences across regions of different levels of urbanicity and disappeared after adjusting for these parameters. However, income, education, and sex were identified as important determinants of disease prevalence and medication use.

The first part of this study describes the burden of musculoskeletal disease in Austria in 2006/2007. Austria was chosen as a study frame for 2 reasons. First, it is a good representation of industrialized nations such as the US or the EU member states. Austria is a federal, parliamentary, democratic republic in Central Europe with a population of 8.3 million people. It is a member of the United Nations and the EU, and is a founding member of the OECD. According to the International Monetary Fund, Austria is the twelfth-richest country in the world per gross domestic product. Second, Austria was the first country to use the EHIS, a highly developed tool created by the EU that will be used regularly and systematically to gather data on the health status of the 500 million inhabitants of the 27 EU member states.

This study used data from this validated instrument and found that two-fifths of the population of Austria have musculoskeletal disease. Earlier studies have shown that musculoskeletal diseases have a grave effect on productivity and economic growth, and affect quality of life and functioning even more than the more “popular” chronic diseases such as cardiovascular or pulmonary disease (19). This effect can add up to almost 3% of the gross national product in developed countries, or approximately 10 billion purchasing power parity dollars in the case of Austria, mostly because of disability and the related loss in productivity (3). No data are available for personal costs for patients with musculoskeletal disease, but it is very likely that these costs will further inflate the abovementioned estimate of $10 billion. Current estimates are that up to 20% of primary health care consultations are due to musculoskeletal disease (20, 21). The data for the prevalence of musculoskeletal disease from this study are well within the range of numbers seen in prior studies in similar countries, further supporting the external validity of the findings at hand (4, 5).

More importantly than describing the sizeable burden of musculoskeletal disease alone, this study aimed at assessing its potential determinants and thus identifying potential approaches to reduce it. Urbanization and urbanicity, or the process of transitioning to and the prevalent state of living in urban areas, have been identified as major contributors to population health in industrialized and developing nations (11–13, 22). More than one-half of the world's population lives in urban areas today, compared with only 40% 25 years ago, according to the WHO (online at http://esa.un.org/unup/). Urbanicity is a surrogate of a conglomerate of numerous risk factors for diseases such as poverty, environmental factors, socioeconomic factors, or simply logistic problems (22). The sum of the detrimental effects of these risk factors is referred to as “urban health penalty.” Childhood populations are particularly susceptible to such determinants of health (23). Austria is a good sample frame to assess the effects of urbanicity on health in industrialized countries since it includes both densely populated urban centers and extensive stretches of rural areas. The assessment of the 14,507 participants of the ATHIS showed that arthritis and osteoporosis were associated with urbanicity, but not spinal conditions. Interestingly, disease prevalence decreased with increasing levels of urbanicity, which is somewhat contrary to the principle of the “urban health penalty.” A potential explanation for this is that the term “urban penalty” was coined in developing countries, in which urbanicity is usually strongly associated with poverty and poor living conditions such as overcrowding and slum formation. In industrialized countries, urbanicity has a number of detrimental effects, such as pollution and stress, but also offers benefits such as better and more specialized health care. However, the final regression model showed no effect of urbanicity on the burden of disease after adjustment for socioeconomic status and demographic variables, showing that not urbanicity itself, but the higher density of wealthier, better educated, and younger people in urbanized centers results in better average musculoskeletal health.

Prevalences for musculoskeletal disease were associated with socioeconomic status, as demonstrated in other studies (24–27). Arthritis rates were lower in the well educated and the well compensated. Spinal conditions, in turn, occurred independent of income level, but were associated with education. Interestingly, osteoporosis revealed an association with both education and income. The strength of these associations was considerable in some cases, with up to 10% and more differences in odds. The fact that education showed a stronger effect than income on all 3 prevalences suggests the use of education as a method of reducing the burden of disease. Such an approach would be promising, since it can be implemented on a large scale, and education can be changed more readily than level of income. Medication use showed even stronger associations with socioeconomic status. With the exemption of osteoporosis, medication use was less in the well-educated and well-earning participants. A potential reason for this could be that more affluent patients have access to treatment options beyond medication. Medication use for osteoporosis was entirely independent of socioeconomic status, suggesting a sufficient level of provision of required pharmaceutical treatment for this disease.

Age and sex were included as demographic variables for 2 reasons. First, age is an important risk factor for musculoskeletal disease and was added to adjust for confounding (28). The same is true for sex, but including this parameter was also important to test for sex inequality. As expected, age was associated with the prevalence of arthritis, spinal conditions, and osteoporosis. Interestingly, there was a strong association of sex with arthritis and osteoporosis, but no association with spinal conditions. The question remains as to what extent these sex effects are biologic and whether they could be partially reversed or averted. Interestingly, there was a strong effect of sex on pain medication used for spinal conditions, but no effect on pain medication used for arthritis, which is the opposite of what was seen for the prevalences of these diseases. Therefore, it seems that women have more arthritis than men, but the use of medication for arthritis is not different among men and women. The opposite is true for spinal conditions for which the prevalence is not different across sex, but women are more likely to take medication for this problem. Whether this is a true biologic effect or a statistical artifact is unclear at this point, but the substantial sample size strongly suggests the former to be the case, and we found no evidence for confounding by socioeconomic status.

The multivariate model that included urbanicity, socioeconomic status, age, and sex produced a number of interesting results. For arthritis prevalence, it showed that the effect of urbanicity was confounded by socioeconomic status, and adjusting for socioeconomic status dissolved the influence of urbanicity on arthritis prevalence. This is a valuable finding because it clearly defines disease determinants that can be monitored or even controlled in future health policy approaches and dampens the considerable impact of arthritis on population health and the economy. A similar effect was seen for osteoporosis, but is less surprising since it is known that age and female sex are the major risk factors for this disease in the studied population. The effects of urbanicity and socioeconomic status on spinal conditions seem to be independent from each other since the OR for the effect of urbanicity barely changed after inclusion of socioeconomic status and demographics into the regression model.

In summary, this study suggests that two-fifths of the population of Austria has musculoskeletal disease. Given the known and sizeable impact of musculoskeletal diseases on productivity and economic growth, such numbers should spur policymakers and public health officials into action (19). Identifying determinants of musculoskeletal disease is a first and important step in reducing their burden. The first impression in this study was that there is an effect of urbanicity on musculoskeletal health, but the data from the ATHIS survey show that the burden of musculoskeletal disease is determined by socioeconomic status and demographic variables, which are differentially distributed across regions of different levels of urbanicity. Our findings suggest that targeted interventions, such as education or support of high-risk individuals, could reduce the burden of musculoskeletal disease by up to 10%. This, in turn, would reduce disability and productivity losses and increase functioning and economic growth.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Vavken had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Vavken, Dorotka.

Acquisition of data. Vavken.

Analysis and interpretation of data. Vavken, Dorotka.

Ancillary