With population ageing, the most prevalent types of diseases shift from acute infectious to chronic non-communicable diseases, such as chronic musculoskeletal conditions. In the most recent Brazilian National Household Survey, around 80% of Brazilians aged 60years or more reported having at least one chronic non-communicable disease, with chronic musculoskeletal disorders being the most prevalent group of diseases (including spine problems, osteoarthritis and rheumatoid arthritis) [4].

Pain is the primary complaint of individuals with chronic musculoskeletal disorders [5], and it is particularly important in the elderly because of its impact on quality of life, independence and social participation. The economic burden of musculoskeletal pain is enormous, being only lower to that caused by cardiovascular disease [6]. Most individuals will present with “nonspecific pain”, a condition in which it is not possible to identify a single specific cause for the pain, even when pain is restricted to one location (e.g. lower back) [7]. This situation is challenging to healthcare practitioners since it requires a more complex approach to diagnosis and treatment. Conversely, a specific diagnosis is possible for some chronic musculoskeletal disorders, such as osteoarthritis. However, regardless of whether a specific diagnosis is possible, the presence of co-morbidities in the elderly makes pain management even more challenging.

The full text of two studies [26,30] could not be retrieved and relevant data were extracted from their abstracts. One of the included studies reported on the results of the Brazilian National Household Surveys conducted in 1998 [16]a], 2003 [16]b] and 2008 [16]c]. Results from each of these surveys were considered separately in this review, totalizing 25 included studies. The included studies reported on a total of 116,091 elderly Brazilians, with individual samples varying from 25 [25] to 41,269 [16]c] participants.

Brazil is a Federative Republic made up of 26 Federation Units called States and the Federal District (Brasilia). Except for the Brazilian National Household Surveys, which recruited participants from all Federation Units [16]a, [16]b, [16]c], the included studies were conducted in seven different States: Bahia [12,13,21,25], Goiás [28], Minas Gerais [14,18,23], Paraná [5,11,20,29,30], Rio Grande do Sul [24,26], Santa Catarina [17,27] and São Paulo [9,10,15,19,22]. Eight studies were conducted in capital cities with population ranging from 421,000 (Florianópolis [27]) to 11.3 million inhabitants (São Paulo [19,22]) [31].

Most study samples included community-dwelling elders, predominantly women in the age group of 60 to 69years old. Three studies [10,12,27] recruited elders seeking care from a healthcare practitioner, and one [13] recruited institutionalized elders. One study recruited both institutionalized elders and elders seeking care in an Emergency Department [20].

The procedure used for data collection and the measure of prevalence of the included studies are listed in Table1. One study [12] collected prevalence data from medical records. Two studies [13,26] used validated questionnaires to evaluate the presence and quality of chronic pain. In six studies the musculoskeletal diagnosis was given or confirmed by a healthcare practitioner after clinical or radiological examination [9,10,20,22,23,27]. In all other studies pain was evaluated by single questions about its presence, location and intensity [5,11,14-19,21,24,25,28-30].

According to the study reports, chronic pain was considered pain that persisted for six weeks or more [18], seven weeks or more [24], three months or more [22], or six months or more [5,11,21,28-30]. In 12 of the included studies, authors did not make clear what was their definition for chronic symptoms [12-17,19,23,25,26].

Most studies (72.0%) investigated the report of present musculoskeletal disorders [9-13,15,16]a,[17,19-22,24,25,27-30], whereas other studies provided prevalence data based on recall periods that ranged from one week [23,26] to lifetime [14,16]b, [16]c, [18]. Six studies [9,14,16]a, [16]b, [16]c, [18] allowed the response to be reported by proxy (either another household member or caregiver).

Eight studies (32.0%) were of high methodological quality [9,10,15,16]a,[21-24]. Only one study fulfilled all the quality criteria [23]. Quality ratings for each study are described in Table2.

The most representative prevalence estimates come from the Brazilian National Household Surveys conducted in 1998, 2003 and 2008, which present data from over 105,200 community-dwelling elders living in every Brazilian State and the Federal District. The use of such broad samples are important in epidemiological research of continental-sized countries like Brazil, where the socio-demographic characteristics of the population are largely variable across the national territory; for example, the Human Development Index (HDI) of Bahia is 0.59, whereas the HDI of Rio Grande do Sul is 0.75 [35].

Interestingly, the prevalence of chronic musculoskeletal disorders has decreased between the Brazilian National Household Surveys conducted in 1998 and 2003 and between the surveys conducted in 1998 and 2008, but not between the latest two occasions of the national survey. This has been attributed to changes in how prevalence was measured in these surveys [16].

The definitions of chronic musculoskeletal pain were largely variable among the included studies. According to the International Association for the Study of Pain (IASP), chronic pain is defined as an episode of pain of at least six months [6]. Specifically for chronic low back pain, the most recent clinical practice guidelines define as chronic an episode of low back pain of at least 12weeks [36]. In most studies included in this review, the definitions of chronic pain did not reflect these recommendations or the authors did not offer sufficient information to judge. The inconsistency among definitions is a problem given that it limits the interpretation and comparison among study results on this topic.

Two previous systematic reviews investigated the prevalence of low back pain in the elderly [37,38], but none of them included the studies conducted in Brazil. Bressler et al. [37] found prevalence estimates for low back pain ranging from 12.8% to 49.0% among community-dwelling elders. According to our results, the prevalence of low back pain in elderly Brazilians ranged from 5.1% to 65.2%. This large variation in prevalence estimates may be due to a number of factors, which include different definitions of chronic symptoms, recall bias and proxy reporting.

The review of Dionne et al. [38] reported the presence of a linear relationship between severe low back pain and age, but not between benign low back pain and age. The positive association between chronic musculoskeletal disorders and age was also reported in two studies included in the present review [11,27]. In the study of Dellaroza et al. [11], this association was present among elders reporting low levels of pain.

Low back pain is currently listed as the most prevalent musculoskeletal disorder among adults in the world [39]. Nevertheless, our results indicate that the prevalence of low back pain is similar to the prevalence of lower limb pain in elderly Brazilians. This finding is comparable to that of Urwin et al. [40], who investigated the prevalence of musculoskeletal disorders in 5,000 individuals from Manchester (UK). The authors found that low back pain was the most prevalent musculoskeletal disorder in individuals younger than 65years old, whereas knee pain was the most prevalent condition in those aged 65years or more, with a peak in women aged 75years and older [40].

It is possible that the differences in self-reported prevalence estimates among the various age groups would reflect extrinsic factors not related to an actual dissimilar distribution of chronic musculoskeletal pain. One factor would be related to the disability (or its perception by the individual) associated with chronic pain. If this is the case, low back pain may be the most prevalent musculoskeletal disorder among elders, but lower limb pain is more frequently reported by them given its greater impact on function, including gait impairments and increased risk of falls. It is also possible that elders consider low back pain as a trivial and less important condition because they may have experienced many low back pain episodes throughout the lifespan that did not lead to any serious consequences to their health.

The presence of co-morbidities, which are common in elders, can also affect their perception of pain. Dellaroza et al. [11] found a significant increase in the report of chronic musculoskeletal pain among depressive elders. A number of recently published studies provide evidence to support the relationship between chronic musculoskeletal pain and depression [41-43]. Interestingly, our findings reflect the importance not only of co-morbidities that directly affect the perception of pain (i.e. depression, anxiety, cognitive deficit), but also co-morbidities that may impact the perception or the report of pain in elders by indirect mechanisms. Some co-morbidities found to be associated with chronic musculoskeletal pain in this review include cardiovascular diseases and diabetes (Table3).

The diagnosis of osteoarthritis is strongly associated with ageing, irrespective of the location (small joints or large weight bearing joints) or gender [44]. Among the specific musculoskeletal diagnoses investigated by the studies included in this review, the broad group of arthritis and rheumatism (including osteoarthritis) was the most prevalent, followed by the diagnosis of herniated disc. In 2004, the World Health Organization (WHO) estimated that over 150 million individuals had osteoarthritis in the world and that this condition was the fifth and ninth cause of years lost due to disability (YLD) in low/middle-income and high-income countries, respectively [45]. Moreover, recent data indicate a staggering 30% increase in the prevalence of this condition in one decade [46]. This large increase is due to a combination of factors, which include ageing of the population, rising prevalence of risk factors (i.e. obesity), and the increased use of imaging [46,47].

We found higher self-reported prevalence rates of chronic musculoskeletal disorders among elderly women. This result is in line with the literature and may be related to fact that women are better at perceiving their physical signs and symptoms than men, and to the knowledge acquired from their role as the family caregiver [48]. Additionally, women may have a higher risk of developing musculoskeletal problems due to anatomo-functional particularities such as shorter height, lower muscle mass and bone mineral density, increased joint laxity and lower degree of adaptation to physical effort [24,49].

The number of epidemiological studies investigating the prevalence of chronic musculoskeletal disorders among elders is still limited, particularly in developing and low-income countries. This reflects the socioeconomic demand for research focusing on the working population. Along with the limited number of studies targeting the elderly population, the low quality of the existing studies makes the interpretation of the evidence still more difficult; for example, only one third of the studies included in this review were of high quality.