Health related quality of life in multiple musculoskeletal diseases: SF-36 and EQ-5D in the DMC3 study
- Correspondence to:
Dr H Susan J Picavet
Department for Prevention and Health Services Research (PZO, pb 101), National Institute of Public Health and the Environment, PO Box 1, 3720 BA Bilthoven, Netherlands;
- Accepted 26 July 2003
Objective: To examine the health related quality of life of persons with one or more self reported musculoskeletal diseases, as measured by the short form 36 item health status survey (SF-36) and the Euroqol questionnaire (EQ-5D).
Methods: A sample of Dutch inhabitants aged 25 years or more (n = 3664) participated in a questionnaire survey. Twelve lay descriptions of common musculoskeletal diseases were presented and the subjects were asked whether they had ever been told by a physician that they had any of these. Their responses were used to assess the prevalence of these conditions. Commonly used scores of SF-36 and descriptive scores from EQ-5D are presented, along with standardised differences between disease groups and the general population.
Results: Subjects with musculoskeletal diseases had significantly lower scores on all SF-36 dimensions than those without musculoskeletal disease, especially for physical functioning (SF-36 score (SE), 75.2 (0.5) v 87.8 (0.5)); role limitations caused by physical problems (67.1 (0.9) v 85.8 (0.8)); and bodily pain (68.5 (0.5) v 84.1 (0.5)). The worst health related quality of life patterns were found for osteoarthritis of the hip, osteoporosis, rheumatoid arthritis, and fibromyalgia. Those with multiple musculoskeletal diseases had the poorest health related quality of life. Similar results were found for EQ-5D.
Conclusions: All musculoskeletal diseases involve pain and reduced physical function. The coexistence of musculoskeletal diseases should be taken into account in research and clinical practice because of its high prevalence and its substantial impact on health related quality of life.
- DMC3, Dutch population based musculoskeletal complaints and consequences cohort study
- EQ-5D, Euroqol five item questionnaire for measuring health related quality of life
- SF-36, Medical Outcomes Study short form 36 item health status survey
The goal of the Bone and Joint Decade 2000–2010 is to improve the health related quality of life for people with musculoskeletal disorders throughout the world.1 The Decade is a multidisciplinary global campaign to implement and promote initiatives in all parts of the world and is endorsed by many international organisations (for example the World Health Organisation and the United Nations) and national governments (see also www.boneandjointdecade.org).
Many instruments are available for measuring health related quality of life. The medical outcomes study short form 36 item health status survey questionnaire (SF-36) and the Euroqol five item questionnaire for measuring health related quality of life (EQ-5D) are two of the most commonly used generic (that is, not disease specific) measures2 used to quantify the health related quality of life in people with musculoskeletal disorders.3–24 Studies employing the SF-36 have been undertaken in patients with chronic back disorders,3–6 arthritis,7 osteoarthritis,8 rheumatoid arthritis,9–14 spinal problems,15 musculoskeletal diseases in general,5,16 and several specific musculoskeletal disorders.16,17 The EQ-5D has been used for back disorders,18 osteoarthritis of the knee,19 rheumatoid arthritis,20 and several musculoskeletal diseases.21 Some studies have analysed both SF-36 and EQ-5D in patients with back disorders,22 arthritis,23 and rheumatoid arthritis.24
Most of these studies focused on only one musculoskeletal disease, but comorbidity of musculoskeletal disorders is common.25 In addition, comparison between studies is often limited owing to differences in study design, case definition and selection, age group, presentation of the data, and probably also language and culture. Data on health related quality of life in patients with different musculoskeletal diseases should preferably be based on a single large dataset and should take into account the coexistence of musculoskeletal diseases. In this paper we present data on health related quality of data (using both SF-36 and EQ-5D) for 12 different self reported musculoskeletal diseases as assessed in a population based survey in the Netherlands. We pay special attention to the effect of multiple musculoskeletal diseases on health related quality of life.
The data were from the Dutch, population based, musculoskeletal complaints and consequences cohort study (DMC3).
The Dutch population of 1998 consisted of over 15 million inhabitants, of whom over 10 million were aged 25 years or more. A random sample of 8000 persons aged 25 years or more, stratified by 10 year age bands and sex (numbers of equal size per age–sex band), was taken from the population register of 1998, identical to general surveys of Statistics Netherlands.26 In September to December of 1998 all questionnaires were sent with a hand signed introduction letter indicating the importance of participation. The net response of the DMC3 study after two reminders (after three and six weeks) was 46.9% (n = 3664). This was calculated by dividing the number of respondents by the number of those actually approached, excluding those who were known to be deceased or whose addresses were unknown. Further information on non-response is given elsewhere.26
We used a 28 page full colour questionnaire consisting of general questions and health questions. On the basis of a list of 12 descriptions of common diagnoses of musculoskeletal diseases, subjects were asked to indicate whether or not a physician ever told them they had any of the diseases. The list was preceded by an introductory text as follows: “There are many diseases of the musculoskeletal system. Some are common, some are rare. Please indicate whether a physician or medical specialist ever told you that you have one or more of the following the diseases.” The descriptions used are given in tables 2 and 3. The description in the questionnaire sometimes differed—thus RSI (repetitive strain injury) also included “for example, a computer arm”; the term osteoarthritis was also accompanied by “wear and tear”; osteoporosis was accompanied by the non-medical term “decalcification of the bones”; and for tendinitis or capsulitis we used “inflammation or condition of tendon or joint capsule.” For this survey we have analysed the test–retest reliability of the self reported diseases and these were acceptable.25 The survey did not include a validity measure such as a comparison with a diagnosis based on physical examination.
Dutch versions of the SF-3627,28 and the EQ-5D29 were used. The SF-36 consists of 36 items that are employed to calculate scores on eight dimensions: physical functioning, role limitation due to physical health problems, bodily pain, general health, vitality, social functioning, role limitation due to emotional health problems, and mental health. Scores had a range of between 0 and 100, with a higher score indicating a better health related quality of life.
The EQ-5D consists of five questions with three response categories. The questions involve the following dimensions: mobility, self care, usual activities, pain, and anxiety/depression. The results of the EQ-5D are expressed as the percentage of subjects with moderate or major problems (any problem).30
To present estimations for the Dutch population, weighting factors were used to make the distribution by age, sex, region, and marital status the same as that in the Netherlands in 1998. We present the health related quality of life scores for 12 different disease groups, eight SF36 dimensions, and five EQ-5D dimensions. These data are also corrected for differences by age. It is not feasible to present statistical tests for every disease by disease comparison. We therefore present the standard errors so that readers can judge for themselves whether a difference is large or not. The difference between two groups reaches statistical significance (at the 5% level) if it is larger than 1.96 times the square root of the sum of the squared standard errors of both groups.
The methods of presenting the results of SF-36 (a score between 0 and 100) and EQ-5D (the proportion with problems) are not directly comparable. To compare the results of these two health related quality of life measures, we also calculated a standardised difference score—that is, the difference between the subject’s score and the weighted score of the general population, divided by the standard deviation of the unweighted score of the general population. This standardised score (the z value or normal score) is a rescaled score with a population average of 0 and a standard deviation of 1. These standardised difference scores are comparable between dimensions and between SF-36 and EQ-5D.
All analyses of data were done using SAS version 6.12.
Table 1 gives the scores of the SF-36 and EQ-5D dimensions in the DMC3 population. The numbers of subjects missing were slightly greater for the SF-36 than for the EQ-5D—the percentages varying from 1.6% (usual activities, EQ-5D) to 8.6% (role limitations from emotional health problems, SF-36). The internal consistency for the different SF-36 subscales was acceptable to good, varying from 0.64 (social functioning) to 0.92 (physical functioning).
The scores from the health related quality of life measures for the different musculoskeletal diseases are presented in table 2 (SF-36) and table 3 (EQ-5D). For all musculoskeletal diseases and all quality of life dimensions it was found that having the disease was associated with a worse health related quality of life. Subjects with any of the 12 musculoskeletal diseases had significantly lower scores on all SF-36 dimensions than those without musculoskeletal disease, especially for physical functioning (SF-36 score (SE), 75.2 (0.5) v 87.8 (0.5)), role limitations from physical problems (67.1 (0.9) v 85.8 (0.8)), and bodily pain (68.5 (0.6) v 84.1 (0.5)). Those reporting a musculoskeletal disease also reported more health problems on the EQ-5D dimensions than those without a musculoskeletal disease—for example, for mobility (29.9% v 10.5%), pain/discomfort (62.5% v 31.2%), and usual activities (34.5% v 12.4%).
With an increasing number of musculoskeletal conditions the health related quality of life deteriorated (fig 1). In fig 1 both the ordinary scores and the standardised scores of SF36 and EQ-5D are given. The standardised scores show that the differences in health related quality of life between subjects with more than one musculoskeletal disease and those without musculoskeletal diseases were similar for SF36 and EQ-5D. The quality of life patterns for five different musculoskeletal diseases expressed as standardised scores (the difference in the number of standard deviations from the population mean) are shown in fig 2. For these diseases the total pattern is shown and also the patterns for subjects with only the particular disease and those with at least one other musculoskeletal disease. Some of these patterns were now based on a very small number of cases, especially, rheumatoid arthritis only (n = 23), and fibromyalgia only (n = 9), so the results give only an indication of the patterns.
In general, the health related quality of life scores for subjects with coexistent musculoskeletal disorders were worse than those with only one specific disease. The patterns were, however, similar.
The dimensions typically affected by musculoskeletal diseases were physical functioning and pain on the SF36, and the dimensions “mobility” and “pain” on the EQ-5D. The diseases with the worst health related quality of life for those dimensions were: osteoarthritis of the knee or hip, rheumatoid arthritis, other types of chronic arthritis, osteoporosis, and fibromyalgia. The diseases with the least severe scores on these dimensions were epicondylitis, whiplash injury, repetitive strain injury, and tendinitis and capsulitis.
For the health related quality of life dimensions involving mental health problems, most musculoskeletal diseases did not score lower than the general population. These dimensions included vitality, role limitation due to emotional problems, and mental health on the SF36, and the dimension anxiety/depression on the EQ-5D. Two exceptions were fibromyalgia (for all these dimensions) and rheumatoid arthritis (only a low score on vitality).
The results of this study show a worse health related quality of life in people with musculoskeletal diseases than in the general population, typically in the areas of pain, physical functioning or mobility, role limitation due to physical health problems, and usual activities. The worst quality of life patterns were found for osteoarthritis of the hip or knee, osteoporosis, rheumatoid arthritis, and fibromyalgia. Health related quality of life scores were lowest among those with multiple musculoskeletal diseases. The results were similar for both SF-36 and EQ-5D.
The specific and substantial impact of musculoskeletal diseases on health related quality of life has already been shown for several disorders. Compared with other chronic diseases, patients with musculoskeletal disorders usually report the lowest health related quality of life.3,5,12,16,23 Comparison of scores between studies is difficult owing to differences in case definition and selection, comorbidity, age, presentation of the data, and probably also language and culture. We will present some comparisons with other studies, taking a few of these methodological differences into account and focusing on SF-36 data for osteoarthritis of the knee and rheumatoid arthritis (table 4).
Our SF-36 scores are more favourable than data from clinical samples.8,24 This is not unexpected because selection of patients in the general population will result in a less severe patient population than selection in hospitals or outpatient clinics. Our SF-36 scores were slightly lower than in self reported disease groups from a general Australian population sample.7 This may reflect a difference in the perception of the disease category between Australian and Dutch subjects.
The strength of our study is the assessment of multiple musculoskeletal diseases. This reveals the important influence of comorbidity on the health related quality of life. We have found no other examples of studies that took the coexistence of several musculoskeletal diseases into account. In the future this factor should be considered during the design of studies and in descriptions of health related quality of life. Other chronic diseases apart from musculoskeletal disease—for example, cardiovascular diseases, cancer, chronic obstructive pulmonary disease, or diabetes—will probably also have a substantial impact on health related quality of life among individuals with musculoskeletal disorders.31 Future studies should pay attention to this type of comorbidity as well.
In presenting health related quality of life data we employed both the commonly used scores for quality of life measures and the standardised difference from the population means. Standardised SF-36 scores have been used by others as well,2,7,11 but their use is not yet common. Some studies have shown deviation from the population norms without dividing by the standard deviation.15,17 We found no example of the use of standardised scores for the EQ-5D. The advantage of the use of standardised scores is that it gives a direct picture of reductions in health related quality of life among disease groups. The choice of population means can, however, be a problem. Some studies have used previously published data on the SF-36,11,15,17 but population norms may depend on the method of data collection (by interview or by questionnaire).32 In addition, population means will differ by country, culture, or language, which reduces the validity of international comparisons even further. However, considering the health related quality of life of chronic diseases in its cultural environment can be viewed as an advantage.
Although the standard SF-36 subscale scores are usually presented as means with standard deviations or standard errors, that procedure is not entirely correct because the scores are categorical—for example, the role–physical score has only the following possible standardised scores: 0, 25, 50, 75, and 100. Alternative measures include the median to describe a group average or to present the percentage scoring below a certain cut off point.33 As far as we are aware, there is no clinically significant cut off point for the SF-36, so for the purposes of this paper we employed the commonly used SF scoring method.
Limitations of our data should be taken into account in their interpretation.
First, self reporting of data has obvious limitations, especially for musculoskeletal diseases. We attempted to exclude self diagnosed or imagined diseases by asking respondents to indicate whether or not a physician had ever told them that they had the disease. It is still possible, however, that diseases are reported that were not diagnosed by a physician. In addition there are also bound to be some undiagnosed diseases. There are sparse data on the reliability and validity of self reports of musculoskeletal disease. Their validity is difficult to assess and the existing data suggest that it is poor.34,35 The strong association between self reported musculoskeletal diseases and a reduced health related quality of life can, however, be interpreted as a indication of validity.
Another limitation of our study is the relatively high rate of non-response. However, on the basis of the general characteristics of the population register, there were no important differences between responders and non-responders, nor did responders in the DMC3 study—a postal survey—differ from responder in an interview survey.26
We employed two commonly used generic health related quality of life measures. Arguments for choosing between SF-36 (eight dimensions) and EQ-5D (five dimensions) include measurement characteristics, the coverage of a broad field of health related quality of life dimensions, quality issues, international acceptance and use of a measure, and the purpose of a specific research project. Advantages of the EQ-5D (five dimensions) include its brevity and simplicity, while the advantages of the SF-36 include its broader coverage, but this is only an advantage if all these dimensions are relevant for a particular research question. The internal consistency of the SF-36 dimensions were acceptable and similar to other data.3,36 The only exception is the dimension “social functioning,” which had a low internal consistency in our study (0.64) compared with others (for example, 0.8336 and 0.803).
If a reduction in questionnaire length is an issue in a musculoskeletal disease study, then the EQ-5D is a reasonable option for a generic health measure, because it covers the most important health related quality of life dimensions for musculoskeletal diseases.
All musculoskeletal diseases involve pain and reduced physical functioning. The coexistence of more than one musculoskeletal disease is important to recognise because it is relatively common and has a substantial impact on health related quality of life.
The Dutch population based musculoskeletal complaints and consequences cohort (DMC3 study) was supported by the Ministry of Health, Welfare and Sport of the Netherlands and the National Institute of Public Health and the Environment and was carried out in collaboration with Statistics Netherlands.