Background

According to the United Nations (UN), the proportion of older people (i.e. aged 60 and over) will triple over the next 40 years and will account for more than 20% of the world’s population by year 2050 [1]. In addition, it is estimated that one in five of the elderly population will be more than 80 years old in 2050. The exponential increase of elderly people is mainly due to a rise in life expectancy, especially in the developing countries. Along with the rise in the life expectancy there is also a rise in the incidence of non-communicable chronic conditions which again leads to increasing morbidity and disability [2]. According to the World Health Organization (WHO), one of the major disabling conditions among the elderly population is musculoskeletal (MSK) disorders [3, 4]. The WHO has specifically identified four major disabling MSK conditions: osteoarthritis (OA), rheumatoid arthritis (RA), osteoporosis (OP), and back pain (BP) [4].

In 1998, the Bone and Joint Decade (BJD) 2000–2010 collaboration was initiated and endorsed by the UN and WHO, with the overall goal to reduce the burden and cost of MSK diseases [5, 6]. In 2003, the WHO’s Global Burden of Disease study and the Bone and Joint Monitoring Project conducted a large report on the burden of MSK disorders through the existing data on the four major MSK conditions (OA, RA, OP, and low back pain (LBP)) [4, 5]. From this report, it is clear that the burden of these major MSK conditions increases with age.

From a health care perspective, the rising proportion and burden of older people demands that health care professionals increase their awareness of the health and disability of this particular population. Accordingly, there is a need to better understand the current magnitude and impact of MSK conditions from this growing population.

The aim of this paper is to estimate the current prevalence of musculoskeletal disorders in the elderly population by conducting a systematic literature review. Specifically, the objective was to estimate the prevalence of non-specific musculoskeletal pain, OA, RA, OP, and BP among older people in developed countries. Any methodological shortcomings will be discussed and future recommendations will be provided.

Methods

Definitions

Musculoskeletal pain in this review refers to the following five overall conditions: 1) non-specific MSK pain in the extremities, 2) RA, 3) OA, 4) OP (either spine or hip or a combination of both), and 5) BP (i.e. neck pain (NP), mid back pain (MBP), and LBP). The older population is defined as people aged 60 and over according to the UN’s cut-off criterion [1]. The term “magnitude” in this review refers to the relative size (i.e. prevalence) of the selected MSK conditions. Hence, the quality of life, cost-of-illness, or social/personal burden of MSK disorders is not included. Developed countries are defined as countries with an advanced economy according to the International Monetary Fund, which includes 35 countries (Additional file 1) [7].

Search design

A systematic literature search was conducted in Pubmed (http://www.pubmed.org) and included studies published between January 1st 2000 and July 1st 2011. The time-period was chosen in order to only include studies published after the WHO reports [3, 4]. Search terms included both free text and MeSH terms and were combined by Boolean terms (AND, OR, NOT) (Additional file 2). The following main terms were included: “musculoskeletal”, “rheumatoid arthritis”, “osteoarthritis”, and “osteoporosis”. The MeSH terms were limited to only include studies containing “epidemiology”, “etiology”, or “diagnosis”. These were again combined with “prevalence”, “cross-sectional studies”. The search was limited by type of papers (review, government publications, technical reports or journal articles), age (MeSH terms: “aged” and “aged, 60 and over”) and finally restricted to English language only. No additional search was conducted. The retrieval of potentially relevant articles was conducted in two phases by one examiner. The first phase focused on identifying relevant studies through the title and abstract. This was followed by retrieval of all full-text articles for further eligibility. As Pubmed adds papers or change MeSH terms retrospectively, the search was repeated after July 1st. The last search was conducted September 1st 2011. No additional searches were conducted, nor were any authors contacted.

Eligibility criteria

Only observational studies from developed countries that reported specific MSK disorders on older people aged 60 and over were included. Thus, studies reporting general MSK pain were excluded. Preferably, the study sample had to represent the general population, but as some individuals may live in nursing homes etc., such studies were also accepted. Table 1 lists the full inclusion and exclusion criteria used in this literature review.

Table 1 Inclusion and exclusion criteria
Full size table

Extraction of information

All core information from the included studies was extracted by an unblinded examiner. The most relevant information were: Article details, study objective(s), study design, method of data collection, sampling method and sample data, disease definition, and outcome data (Table 2). If the included study referred to another reference (i.e. another paper, report, or website) for a more detailed description of the study cohort, then that reference was perused for additional information if it was accessible.

Table 2 List of items extracted from each article
Full size table

Risk of bias assessment

The quality of each study was determined by assessing the risk of bias [8]. Recently, Viswanathan et al. have identified 29 practical and validated items that may be used to evaluate the risk of bias and precision of observational studies [9]. This bank of items covers a range of different study designs and the authors have provided instructions as to what items to use depending on the studies under assessment. Thus, only items related to our main objectives were identified and criteria for each item were defined to fit our main objective (Table 3). The layout of the questionnaire was slightly modified for practical reasons, but no other changes were made. The chosen items focused on selection bias, information bias, and the overall interpretation of each study. Relevant criteria to assist in determining the risk of bias in a study were specified to each item. No validation of the included items was performed.

Table 3 Items chosen to assess risk of bias of the included studies
Full size table

Data analysis

The extracted data was presented in separate tables for each of the included MSK conditions. In studies where the results were only presented graphically, best effort was made to determine the prevalence estimates from the graphs (without decimals). Both total and gender prevalence estimates as well as age related changes were reported when possible. In addition, the attempt was made to present pooled means of prevalence estimates on fairly homogeneous studies.

Results

Search results

In total, 5097 articles were found through the search strategy (Figure 1). Based on either their title or abstract, 185 were subsequently retrieved and reviewed. Of these, 100 articles were rejected, mainly because prevalence estimates on elderly aged 60 and over was not reported or could not be determined (82%) (Additional file 3). Other reasons for exclusions were 1) the studies did not fulfil the inclusion/exclusion criteria (14%) and 2) articles reporting results that were already published in other articles (i.e. duplicate publications) (4%). Thus, in all 85 articles were included in this review.

Figure 1
figure 1

Flow chart of search results.

Full size image

Study characteristics

The included articles were published in 39 different journals of which 4 journals (Spine (26%), Rheumatology (18%), Annals of Rheumatic Diseases (18%), Arthritis & Rheumatism (15%)) accounted for approximately three quarters of all journals. There was an uneven distribution of publications between 2000 and 2011, but with no clear patterns across the decade. The majority of the studies were from Europe (58%) followed by Australasia (21%), North America (18%) and Middle East (4%).

Risk of bias within each study and across studies

Overall, 25% of the studies were determined as having a low risk of bias and 11% were deemed as having a high bias risk (Figure 2 and Additional file 4). Thus, in approximately 65% of the studies it was unclear if risk of bias were either low or high, mainly because it was difficult to determine if the final study sample was truly representative of the target population. The risk of bias for each of the included studies is presented within each of the musculoskeletal conditions.

Figure 2
figure 2

Risk of bias – Summary of all studies.

Full size image

Prevalence of musculoskeletal disorders in the elderly population

A total of 173 different prevalence estimates were extracted from the 85 included studies. The most commonly reported MSK condition (i.e. number of prevalence estimates) was BP (29%), OA and OP (17%), followed by RA (8%), ankle/foot pain (8%), knee pain (6%), hip pain (5%), shoulder pain (5%), hand/wrist pain (3%), and elbow pain (3%).

Prevalence of RA

Rheumatoid arthritis was described in 12 studies with a total of 13 different point prevalence estimates [10–21] (Table 4). Seven (58%) were of low risk of bias [10, 11, 13, 16, 17, 19, 20] and only one study [18] was deemed as being of high risk of bias (Table 4 and Additional file 4).

Table 4 Description of studies on rheumatoid arthritis (RA)
Full size table

The prevalence estimates that were based on clearly defined criteria (typically the 1987 American College of Rheumatology (ACR) criteria [22]) ranged between 0.4% and 2.2%. The prevalence of RA was higher among women. No clear age related differences could be determined, but generally the prevalences were minimal across ages.

Prevalence of OA

Sixteen studies reported prevalence estimates on OA in four different anatomical sites (knee, hand, hip, and lumbar spine) either based on symptomatic findings only, radiographic findings only, or on a combination of both [11, 18, 23–36] (Table 5). Of these studies, five (31%) were judged as being of low risk [11, 23–25, 30] and only one study (6%) of high risk of bias [18] (Table 5 and Additional file 4).

Table 5 Description of studies on osteoarthritis (OA)
Full size table

Lumbar spine OA

Two Japanese studies on lumbar spine radiographic OA, using a higher Kellgren-Lawrence (K-L) grade (≥3), reported point prevalences of 40%-75% in the 60–69 year olds to 80%-90% in the 80+ age group [33, 35].

Hip OA

Only three studies on hip OA were found in this review [18, 23, 37], two studies on symptomatic hip OA [18, 37] and one on combined symptomatic/radiographic hip OA [23]. The self reported hip OA were about three times higher (17-22%) than found through clinical examination (approx. 8%) and more common in women than in men [23]. Combined symptomatic/radiographic hip OA increased from 2% in the 60–64 year olds to 3% in the 75–79 year olds, but then decreased slightly in the 80+ year olds.

Knee OA

Knee OA was reported in 11 studies [11, 18, 23, 25, 27, 28, 30–34] and presented 14 different prevalence estimates (Table 5). The ACR clinical criteria [38] for knee OA was used in two out of three studies on symptomatic knee pain and showed fairly similar prevalence estimates (28-33%).

All studies on radiographic knee OA only (i.e. without reported pain) either used the K-L grade 2 [39, 40] or higher criteria for OA [27, 28, 31–34]. Nevertheless, great variations in point prevalence estimates were reported. For example, in women in their sixties, OA was present in 40% to 57%, and in the seventies it ranged between 54% and 74%. In men, larger differences were found (60s: 4%-35%) and (70s: 18%-51%). Overall, higher OA estimates were reported with increasing age.

For the combined knee OA and reported pain, generally larger gender differences were seen (Table 5) and more variation in age trends were also noted [23, 27, 28, 31, 32]. Painful knee OA increased with age until approximately at age 80+ where a slight decrease was reported in two out of the four studies [11, 18, 25, 30].

Hand OA

Seven studies included data on hand OA [11, 23, 24, 26, 29, 36, 37] with a total of eight prevalence estimates on symptomatic [11, 37], radiographic [26, 29], and combined symptomatic/radiographic hand OA [23, 24, 29, 36] (Table 5).

Regardless of hand OA definitions, women had more OA than men and overall, OA increased with age, although several studies also reported a slight decrease in the oldest age groups.

Five studies reported either symptomatic hand OA only [11, 37] or radiographic hand OA only [24, 26, 29], all with different definitions and age ranges. Nevertheless, similar point prevalences were noted: Approximately 15% of the “younger” elderly population reported symptomatic hand OA. Radiographic hand OA ranged from approximately 56% in the “youngest” elderly men to 100% in the oldest women.

The point prevalence estimates of combined symptomatic/radiographic hand OA ranged from approximately 4% in the “youngest” elderly population to approximately 14% in the oldest people and were therefore less common than radiographic hand OA alone.

Prevalence of OP

Twenty-one studies reported prevalence estimates on OP of which 14 studies measured the bone mineral density (BMD) in five well-defined anatomical areas (lumbar spine/hip, lumbar spine only, hip/femoral neck only, hand, and heel) [33, 35, 41–52]. Seven studies used other definitions and were mostly based on self reported data [12, 18, 53–57] (Table 6). Four studies (19%) were of high risk of bias [18, 47, 51, 54], whereas only two studies (10%) were of low risk of bias [41, 52] (Table 6 and Additional file 4).

Table 6 Description of studies on osteoporosis (OP)
Full size table

Regardless of the anatomical site, a steady increase in OP with increasing age for all types of OP definitions was seen. Generally, OP was two-three times more common in women than in men.

Lumbar spine OP

Eight studies included data on lumbar OP [33, 35, 44, 45, 49–52], all using the WHO BMD T-score of −2.5 SD or less [58], except for two studies [33, 51] (Table 6). While the Spanish and Danish OP age related prevalences in women were similar (ranging 17%-66%), greater age related variations were noted in women in the Asian countries. For example, in South Korean women, markedly higher estimates across ages (51%-61%) were reported by Cui et al. [44] compared to Shin et al. (29%-48%) [50].

Hip or femoral neck OP

Seven studies reported either hip or femoral neck OP [33, 44–46, 49, 51]. Fairly similar results were noted in South Korea and Australia (range: 11%-37% for 60–79 year olds) [44, 45], but the UK and Spanish estimates were slightly lower (range: 7%-15% for 60–74 year olds) [46, 49].

Combined lumbar spine and/or hip OP

Lumbar spine and/or hip OP was reported in five studies [41, 43, 48, 49, 52] which all, except for one study [43], used the WHO bone mineral density (BMD) threshold (T-score) of −2.5 SD or less (Table 6). The prevalence of OP was slightly higher in Danish women [52] (range: 30%-92%) than in Spanish women [49] (range: 23%-49%).

Prevalence of BP

In all, BP 31 studies were included [11, 41, 59–87] of which seven (23%) studies were of low risk of bias [11, 41, 73–75, 78, 80] and three (10%) of high risk of bias [59, 81, 83] (Table 7 and Additional file 4).

Table 7 Description of back pain (BP) and neck pain (NP)
Full size table

Low back pain

Low back pain was reported in 20 studies all with different LBP definitions and with eight different prevalence periods (Table 7) [11, 41, 59, 60, 64, 66, 69, 71, 74, 75, 77–83, 85–87].

The one-month prevalence was the most common prevalence period reported and ranged between 27% and 49%. The lowest estimates were based on more restricted definitions, whereas the larger estimates (47-49%) had less restricted LBP definitions.

Overall, the prevalence estimates increased up to 80 years of age and then dropped slightly after that. With one exception [83], women reported LBP more often than men.

Back pain

Back pain was used in six studies [62, 63, 68, 70, 72, 73] on five different prevalence estimates, all with different BP definitions and with a wide range in prevalence estimates. Thus, one-month BP prevalence ranged between 18% and 29%, and the point prevalence ranged from 27% to 58%. Interestingly, in two studies where 100 year olds were included, the point and one-month BP was roughly the same (27%-29%) [63, 70]. Prevalence estimates were all higher among women, but age-related changes are inconclusive as most studies did not demonstrate any major changes across ages.

Neck pain

Sixteen studies on NP reported six different prevalence periods [41, 61, 67–70, 73, 75–78, 82, 84–87] of which the one-month prevalence was the most commonly used period. No identical NP definitions were used and/or different age intervals were reported, although some definitions and intervals were fairly similar.

Overall, the one year prevalence ranged between 9% and 12% [41, 61, 71, 84]. Greater variations were noted for the three-month prevalence, ranging between 5% [77] and 56% [75] in 65–74 year olds. Of the four one-month prevalence estimates using fairly similar NP definitions, about 23% reported NP [70, 76, 85, 87]. Men reported NP less often than women and in all studies there was a decrease in NP with increasing age, albeit small in some studies.

Mid back pain

Finally, MBP (i.e. thoracic or higher back pain) was reported in three studies [75, 77, 78]. The three-month prevalence was used in two studies, but with different MBP definitions and thus, the prevalence ranged between 2% [77] and 15% [75]. One study showed that pain in the “higher back” was four times more prevalent among women [78].

Prevalence of shoulder pain

Six studies reported five different prevalence periods on shoulder pain [73, 77, 78, 84, 86, 88] and two studies also included upper arm pain using two different prevalence periods [87, 89] (Table 8). Two studies (25%) were rated as having low risk of bias [73, 77, 78] and the rest as having an “unclear” risk of bias (Table 8 and Additional file 4).

Table 8 Description of studies on shoulder pain
Full size table

All studies used different shoulder pain definition and/or different prevalence periods. Nevertheless, in some of the studies with different prevalence periods, the estimates varied only slightly (3-5%) (65–74 year olds, men: 10%-13%; women: 18%-23%) [73, 78, 89]. In three studies where gender estimates were provided, women reported more pain than men [73, 78, 89]. Only one study provided different age intervals, which showed that shoulder pain increased slightly with age.

Prevalence of elbow pain

Elbow pain was reported in four studies [73, 77, 78, 86] and elbow/forearm pain in one study [89], of which three different prevalence periods were used (Table 9). Two studies (40%) were of low risk of bias [73, 77, 78], and the rest being unclear (Table 9 and Additional file 4).

Table 9 Description of studies on elbow pain
Full size table

Different elbow pain definitions were used in each study. Nevertheless, similar estimates were reported for both point and three-month prevalences [73, 78]. Thus, approximately 5% of men and 6%-8% of women reported elbow pain. Elbow pain increased with age [73, 77]. Fewer men reported elbow pain compared to women [73, 78].

Prevalence of hand/wrist pain

Two studies reported hand pain only [73, 87], one study wrist pain only [77], and three studies on combined wrist/hand pain [78, 86, 89] (Table 10). Two studies (33%) were of low risk of bias [73, 77, 78], and the rest were unclear (Table 10 and Additional file 4).

Table 10 Description of studies on wrist and hand pain
Full size table

Wrist and/or hand pain prevalence estimates varied greatly among the different studies. For example, as few as 14% of men aged 75+ [73] and as many as 26% of women aged 60–69 [87] reported hand pain. Also, 2% of men between 65–74 [89] and 22.5% of women (65+) [78] reported wrist/hand pain. Women reported more often wrist and/or hand pain than men [73, 78, 89]. Hand pain increased slightly with age in one study [73], but decreased in the other study [87].

Prevalence of hip pain

Five different prevalence periods on hip pain were reported in nine studies [73, 75, 77, 78, 83, 87, 90–92] (Table 11). Three studies (33%) were considered to be of low risk of bias [73, 75, 78] and only one study (11%) of high risk of bias [83] (Table 11 and Additional file 4).

Table 11 Description of studies on hip pain
Full size table

All nine studies used different hip pain definitions, resulting in a wide prevalence range. For example, the three-month prevalence ranged between 5% and 30% in the elderly aged 65–74 [73, 75, 77]. Six studies reported gender specific prevalence estimates, all of which reported a higher prevalence in women [73, 78, 83, 90–92]. Age related changes were somewhat unclear and only showed small (2-4%) differences across age groups.

Prevalence of knee pain

Eleven studies reported five different prevalence periods on knee pain [27, 73, 77, 78, 83, 86, 87, 91–94] (Table 12). Three studies (27%) were of low risk of bias [73, 78, 94] and one study being of high risk of bias [83] (Table 12 and Additional file 4).

Table 12 Description of studies on knee pain
Full size table

All 11 studies used different pain definitions which resulted in great variations in prevalence estimates. For example, in the 65–74 year olds, the one-year prevalence varied between 26% and 70% in men and between 36% and 71% [91, 92]. Generally, there was an increase in knee pain with increasing age, ranging between 3% and 8% [27, 73, 92, 94]. Some studies reported a slight decrease [91, 93] whereas others found no change with increasing age [77, 87]. Five studies included gender specific prevalences and all showed that more women than men reported knee pain [73, 78, 83, 91, 92].

Prevalence of ankle/foot pain

Nine studies included information on foot pain [73, 75, 78, 87, 92, 95–98], three studies on ankle pain [78, 86, 99], and one study on both ankle/foot pain [77] (Table 13). Of these 12 studies in total, five (42%) were of low risk of bias [73, 75, 78, 96, 98] and only one study was considered being of high risk of bias [97] (Table 13 and Additional file 4).

Table 13 Description of studies on ankle and foot pain
Full size table

Two studies with similar designs and definitions reported that 23%-29% of 60–80 year olds had pain in their feet during the past month [87, 98]. In contrast, two other similar studies on point prevalence showed greater variations (65+ men: 9%-14%; women: 12%-28%) [78, 96]. Otherwise, great variations in prevalence were found, for the same reasons as described under the wrist/hand pain section. In all the studies reporting gender prevalences, women suffered more from ankle and/or foot pain than men [73, 78, 92, 96, 99]. In two studies, foot pain increased with age [73, 75], but dropped in another study [87].

Musculoskeletal co-morbidity

Information on multiple/widespread MSK conditions in the elderly population was extracted from 15 studies [30, 59, 68, 72, 75, 78, 82, 84, 86, 87, 89, 91–93, 100].

In a Danish elderly population (70–120 year olds), concurrent neck and BP was found in 13% of women and 8% of men [68]. The same findings were reported in the USA, where 9% of 65+ year olds had both NP and LBP [82]. Jacobs et al. reported an almost two-fold increase in concurrent joint pain among older people (70 and 77 year olds) with chronic BP (59% and 74% respectively) compared to those without chronic BP [72].

Widespread pain was reported in the study by Natvig et al., where 14-15% of Norwegian people aged 64–86 years had additional MSK pain (from either shoulders, elbows, hands/wrists, upper back, lower back, hips, knees, or ankles/feet) [100]. In Sweden, between 4% and 6% of men aged 65–74 with upper extremity pain also reported either NP, LBP, or lower extremity pain, whereas in women the reported prevalence was about three times higher (15%-17%) [89]. According to Vogt et al., 14% of 70 to 79 year old Americans reported concurrent MSK pain in at least four sites [84]. In the UK, three studies on multiple pain sites showed varying results among 65+ year olds, which may be due to different definitions [91–93]. According to Dawson et al., 11% of the older adults had both hip and knee pain [91]. Croft et al. reported slightly higher estimates (26%-33%) but included the whole body [93]. In the study by Peat et al., 40% had more than one painful joint in the lower extremity [92]. More widespread pain (up to 44 pain sites) was reported by 12%-16% of women and by 7%-13% of men aged 60 and over [87]. In Italy, “polyarticular peripheral joint pain” was reported in 28% in the same age group (65+) [30]. In a Dutch study, multiple MSK pain sites were present in roughly 28% of men and in 46% of women aged 65 and over [78].

Other studies report several MSK pain sites in more than half of the elderly people, which indicates overlapping MSK symptoms [59, 75, 86]. In a South Korean elderly population (65+), more than half reported both upper extremity pain as well as LBP and/or lower extremity pain [59]. Similarly, in an Israeli population of elderly people aged 61 and over, more than half reported LBP, NP, knee and shoulder pain [86]. Furthermore, at least a third of these people also reported other peripheral joint pain sites. Finally, in a Spanish study, people aged 65 and over had on average four MSK pain sites [75]. Unfortunately, it is not possible to determine how many of these suffered from multiple pain sites. Thus, based on these three studies, a high degree of overlapping/concurrent MSK pain sites must be present [59, 75, 86].

In summary

  • The prevalence of MSK conditions remains high even in old age regardless of the type of complaint.

  • Women typically report problems more often than men, regardless of the MSK condition.

    The prevalence of MSK complaints typically drops slightly in the oldest age group (i.e. 80+ year olds), except for OP where all studies report an age related increase.

  • Widespread/concurrent MSK pain is very common among elderly people, affecting every second or third elderly person.

Discussion

Summary of evidence

In this review a great variation in prevalence of MSK disorders in older people were found. The most likely reasons for these differences are: 1) different pain definitions, 2) different prevalence periods, 3) different age intervals, and 4) the prevalence estimates were either divided by gender or only reported as a total prevalence estimate. Thus, it is impossible to determine any overall estimates on the prevalence of MSK problems in the elderly population.

Nevertheless, some general observations can be drawn from this review that needs to be discussed. Musculoskeletal disorders remain prevalent in the elderly population. Especially, OA is very common among elderly people, followed by knee pain, BP, and for women also OP. Pain mechanisms in the older population are poorly understood, but it is generally believed that pain at younger ages continues in the older ages [101]. Thus, pain in the elderly should be regarded as a continuum of pain from earlier years [101].

Women tend to report MSK pain significantly more often than men in almost all studies. This gender difference in pain reporting is well known, but the reason for this is probably multifactorial with both biological and psychosocial underlying mechanisms. These different pain mechanisms are beyond the scope of this paper to discuss in detail, but are presented in a review by Fillinghim et al. [102].

There is a general trend that prevalence estimates either remain fairly constant with increasing age or that they drop slightly in the oldest people, typically from 80 years of age and onwards. An exception from this is OP, where a steady increase is reported with increasing age.

There are several potential explanations for this decline in pain reporting with age. It may simply be a general birth cohort effect which may reflect both cultural and public health related differences between for example 40 year olds and 80 year olds [103]. This potential cohort effect may be more pronounced in cross-sectional studies, which were the only included studies in this review. A parallel to this may be that pain is accepted by the elderly as part of becoming old [104]. In other words, pain becomes a natural part of their life and therefore become less disturbing or simply ignored. It is also known that pressure pain decreases with age [105]. Finally, a decline in pain prevalences in the oldest old could be explained by a “survival of the fittest” phenomenon [103]. However, MSK pain itself does not lead to premature mortality per se [106–108]. Furthermore, this “biological elite” phenomenon is probably slowly diminishing as health and living standards in the World is generally improving and thus, more people are living longer and generally at better health

Finally, there is a considerable degree of overlapping MSK symptoms as approximately every second or third elderly have widespread MSK pain. This trend is most likely part of a continuum from widespread pain at younger ages as previously mentioned [101].

Comparisons with other reviews

To our knowledge, no previous systematic literature reviews on a broader range of MSK conditions in elderly populations exist. However, a few reviews on some of our MSK conditions in the elderly populations were identified. Woolf and Pfleger reported high prevalence estimates in the elderly people for OA, RA, OP, and LBP in the developed countries [4]. In all four MSK diseases, the same age related increase in prevalence was found in their review, except for LBP where it remained fairly constant.

A literature review on LBP before 2000 found only 12 prevalence estimates specifically on elderly populations, but the authors were unable to make any general estimates mainly because of the different (or lack of) LBP definitions as well as the varying age intervals [109]. In a more recent LBP review published in 2006 on age related changes, concluded that “benign” LBP decreased with age, but that more severe LBP increased with age [110]. Due to the heterogeneity of these studies and the aim of their review, no attempt was made to provide any general LBP prevalence estimates.

Luime et al. published a review in 2004 on shoulder pain [111]. The point prevalence on subjects <70 ranging 7%-27% was very similar for subjects older than 70 (12-26%), but this may be due to the varying pain definitions.

Dagenais et al. found a steady increase in hip OA with increasing age, ranging from 5% (60–64 year olds) to 14% (85+ year olds), and being more prevalent in women [112].

It is impossible to compare our results with the abovementioned reviews, as they too fail to provide pooled estimates due to the high degree of heterogeneity across the included studies. Nevertheless, a general increase in prevalence with age and a gender difference were reported in all reviews, which is in accordance with our own findings.

Methodological issues

The heterogeneity of pain definitions is already a well known problem, but undoubtedly, researchers have many good reasons for why they use a specific and perhaps unique pain definition. Unfortunately, this makes it impossible to draw any general conclusions based on the currently available literature. However, it would be recommendable if authors would at least report one or two additional standardised measures, such as the questions from the standardised Nordic questionnaire on musculoskeletal pain [113]. Although, journals restrict the sizes of their papers by limiting the number of words or tables and hence, decreasing the amount of information available from the studies, it is becoming more and more common to have supplementary tables published via the publishing journal’s website. Such tables could include valuable information on gender specific and total prevalence estimates for future reviews to calculate pooled prevalence estimates.

It also needs mentioning that nearly twice as many prevalence estimates could have been obtained from 82 additional studies, if only authors had reported age specific estimates. So, just like the standardisation of pain definitions is warranted, standardisation of age interval reporting would also be preferable. This way, more information on age related changes from the current literature could easily have been obtained.

In this review, we found that many authors state that their results are representative of the general population. However, only few actually document this. While many do their best at obtaining a random and representative target sample from the background population, an actual non-response analysis is rarely performed. For this reason, the risk of bias of the majority of the studies (65%) was deemed unclear. Studies were generally judged as having an “unclear” risk of bias because information was missing in the study description. In other words, the external validity of these studies is questionable, which is essential in epidemiological studies. It is therefore important to either report and/or adjust for non-response bias in future studies.

Strengths and limitations of this review

Just like our included studies, our review has also some limitations that need to be addressed. We only included one electronic database (Pubmed) and thus, may have missed some relevant articles. Based on other reviews on similar MSK conditions, who have included other electronic databases (i.e. EMBASE, CINAHL, etc.), we may have missed between zero and 12% potentially relevant articles [109–112]. However, given the large heterogeneity and therefore lack of proper summary prevalence estimates, we doubt any missed articles would have had any major impact on our results. Our search strategy was also limited to the elderly population through MeSH terms. This may have lead to exclusion of some studies if for some reason they were not properly indexed in Pubmed. As only English language articles were included, any articles published in national non-English medical journals are missing in our literature review. Finally, the selection of articles was only conducted by one author, thus, there is a risk of missing potentially relevant articles. According to Edwards et al., an average of 9% of relevant articles may be missed (ranging between 0 and 32%) [114]. Thus, on average we may have missed approximately 8 articles.

The results from the included epidemiological studies must be viewed in light of the quality of these studies which depends on both the internal validity and if the results can be extrapolated to the background population (i.e. the external validity). In this review, the risk of bias rather than the quality of the studies were used as we wished to determine if the results were “believable” and not just if the “reporting” was satisfactory. The risk of bias assessment on randomised clinical trials is also recommended by the Cochrane Collaboration [8] and recently a set of risk of bias items were developed by Viswanathan et al. [9] which allowed us to design an assessment sheet well suited for our needs. However, assessing the risk of bias demands a high degree of judgement, is more time consuming, and may result in greater variability of interpretations of the studies [9, 115]. Therefore, no attempt at adjusting the prevalence estimates based on the risk of bias judgment was made. Instead, we leave it up to the readers to decide on how to utilise our risk of bias judgments.

Because MSK pain may be reported as part of a larger health related publication and because a wide set of MSK conditions were included in our review, it was necessary to have rather broad search strategy. This in turn, resulted in a very large number of hits that had to be perused to seek for any potentially relevant articles. While the search may have been fairly sensitive in catching relevant articles it cannot be considered to be very specific. This becomes clear as less than 4% of the initial search results were retrieved and only 46% of those included. We did not attempt to specify the literature search any further as some of the included articles would have been missed, especially those articles where the reporting of MSK conditions are “secondary” findings.

Another limitation is the choice of only investigating the prevalence of MSK disorders among elderly people and, hence, excluding information on burden and cost-of-illness of these MSK conditions. Clearly, the presence of pain does not reflect how MSK problems affect older people on a daily basis. However, in the 2003 WHO report, Woolf and Pfleger reported that MSK conditions have a major societal impact in terms of reduced work disability, which would affect the “younger” elderly people aged 60–65, and result in an increased use of health care services [4]. Finally, with increasing OP, there is a high risk of fracture incidences. As most MSK conditions remain fairly common in the elderly populations and as the number of elderly people increases in the future, the socioeconomic burden of MSK in the elderly population will also increase. Thus, there will be a further need for health care professionals to deal with chronic MSK conditions among the elderly people.

Future perspectives

This review has looked at the prevalence of a series of musculoskeletal conditions in the elderly population and will serve not only as a reference for future studies, but also as a guide for clinicians in general. Firstly, a larger population of geriatric patients must be expected in the future and thus calls for more attention on developing optimal geriatric patient management protocols. Secondly, it is important for a person to maintain a sufficient functional capacity in order to maintain an active life at older age [3]. In other words, political programmes as well as primary and secondary health care programmes accommodated to the future needs are necessary in order to maintain (or ideally improve) the quality of life in the elderly population.

Conclusions

No overall estimate on the prevalence of MSK problems in the elderly population can be determined due to the heterogeneity of the studies. However, MSK disorders are common in the elderly population and women have more often MSK problems than men. There is a general trend that prevalence estimates either remain fairly constant or increase slightly with increasing age. However, for many MSK conditions, there is a slight decrease among the oldest (80+) people. Finally, many elderly people report multiple MSK pain sites.