Stewart Willians 2015
Stewart Willians 2015
Stewart Willians 2015
collection and analysis, decision to publish, or describe factors associated with back pain prevalence and intensity, and back pain as a de-
preparation of the manuscript. terminant of disability.
Competing Interests: The authors have declared
that no competing interests exist.
Results
Prevalence was highest in the Russian Federation (56%) and lowest in China (22%). In the
pooled multi-country analyses, female sex, lower education, lower wealth and multiple
chronic morbidities were significant in association with past-month back pain (p<0.01).
About 8% of respondents reported that they experienced intense back pain in the previous
month.
Conclusions
Evidence on back pain and its impact on disability is needed in developing countries so that
governments can invest in cost-effective education and rehabilitation to reduce the growing
social and economic burden imposed by this disabling condition.
Introduction
Back pain is a highly prevalent disabling musculoskeletal condition affecting almost everyone
at some time [1]. The biopsychosocial model is the prevailing framework used for understand-
ing, managing and treating back pain. This approach suggests that in addition to biology, psy-
chological, socio-economic, environmental and cultural factors all contribute to the incidence
and persistence of back pain symptoms [2, 3]. Many musculoskeletal conditions start in mid-
dle-age and require interactions with health care providers over many years [1, 4, 5]. Low back
pain, or “back pain”, is a leading cause of activity limitation, work absenteeism and lost produc-
tivity throughout much of the industrialized world–threatening function, mental health and
quality of life [6–8] and inflicting substantial direct and indirect costs on health, social and eco-
nomic systems [1].
Globally back pain causes more disability than any other condition. The 2010 Global Bur-
den of Disease Study ranked low back pain as the condition with the highest number of years
lived with disability (YLDs) and sixth in terms of disability-adjusted life years (DALYs) [8, 9].
In 1990, the global burden of YLDs due to back pain in adults aged 50–69 was 59% in develop-
ing countries, but by 2010 this proportion had increased to 67% [10]. With rapid growth in the
numbers and proportions of older adults in low- and middle-income countries (LMICs) the
back pain burden in older adults in these countries is expected to grow significantly in coming
decades [1, 8, 9, 11].
Back pain is also one of the most common conditions for which patients in high-income
countries seek medical care [12]. Most of the information about back pain has come from de-
veloped countries in Europe, North American and Australasia, making it difficult to draw com-
parisons with developing countries [12–15]. Italian researchers reported back pain prevalence
of 32% in adults aged over 65 years [16] and a study of community-dwelling adults aged 70–79
years in the United States (US) demonstrated back pain prevalence of 36%. A review of the
prevalence of musculoskeletal conditions in adults aged 60 and over in developed countries re-
ported one-month back pain prevalence of between 18% and 29% [17].
The Jerusalem Longitudinal study [18] showed that chronic back pain was prevalent in the
elderly (aged 70 years and over) and that psychosocial factors, female gender and hypertension
were associated with back pain. Association between back pain and older age is also heavily
modified by the severity and intensity of the complaint [19]. Studies conducted in North Amer-
ica, [20, 21] Europe, [12, 22, 23] and Australasia [17] found that the prevalence and intensity of
back pain is associated with individual, psychosocial and occupational factors. In addition to
being older and female [1, 24, 25] modifiable determinants of back pain in developed countries
include smoking, depression, lack of physical activity and abdominal obesity [12, 23, 24, 26–
30]. A Japanese study of men aged 40 years and older demonstrated that back pain had a signif-
icant negative impact on quality of life [31]. European studies provide evidence of inverse asso-
ciation between back pain and socioeconomic factors, such as older age, higher income and
education [32–35].
Although a few studies investigating the determinants of back pain have been conducted in
developing countries, the literature is sparse compared with developed countries. In a commu-
nity-based study of adults in Korea (mean age 56 years) the common determinants were ad-
vancing age and female sex [36]. Studies conducted in Taiwan [37], China [38] and Sri Lanka
[39] have focused on working-age populations. A review of back pain prevalence studies con-
ducted in Sub-Saharan Africa on mostly working-age adults and adolescents, concluded that
back pain prevalence was rising [25].
The perception and reporting of back pain is influenced by individual characteristics, work-
ing conditions, lifestyle, and social, economic, cultural and ethnic factors, as well as the avail-
ability of treatment and rehabilitation options [1, 5]. In some societies and countries there is a
greater awareness of the symptoms and also a greater willingness to report them, while in oth-
ers, back pain is not necessarily associated with disability, but rather seen as a natural conse-
quence of routine physical work or the ageing process itself [1]. Given the fundamental social,
cultural and economic differences between developed and developing countries, it is reasonable
to argue that the antecedents and consequences of back pain are not homogeneous. For exam-
ple, extreme poverty, infectious diseases epidemics, work tasks, family structures, responsibili-
ties, social expectations, geography, health care availability and support may all impact
differently on the perception and reporting of back pain in different contexts and settings.
As a consequence of the major epidemiological and demographic transitions occurring in
emerging economies in all regions of the world, there is now an urgent need to gather accurate
comparable epidemiological data on back pain in older adult populations in developing coun-
tries [11, 29, 36, 40–43]. This research draws attention to the complexity of predisposing, en-
abling and contextual factors that require consideration at the country level [5].
According to the World Health Organization (WHO) one of the most disabling conditions
among the elderly is musculoskeletal disorders, of which back pain is a major contributor [8,
14, 17, 40, 43]. Data collected from adults aged 50 years and older participating in the WHO
Study on global AGEing and adult health (SAGE) were analysed to: measure the prevalence of
back pain; identify risk factors and determinants associated with back pain prevalence and in-
tensity, and describe association between back pain and disability. The purpose is to gain in-
sights into country-level differences in self-reported back pain in a group of socially, culturally,
economically and geographically diverse LMICs.
Methods
Ethics Statement
The SAGE study was approved by the following bodies: the Ethics Review Committee, World
Health Organization; Ethical Committee, Ghana Medical School, Accra, Ghana; Ethics Com-
mittee, OPM (School of Preventive and Social Medicine), Russian Academy of Medical Sci-
ences, Moscow, Russia; Ethics Committee, Shanghai Municipal Centre for Disease Control and
Data Collection
The SAGE Wave 1 is a longitudinal study that provides the baseline round of data for national-
ly representative samples of adults aged 50 years and over in China, Ghana, India, Mexico, the
Russian Federation and South Africa. Cross sectional data from SAGE Wave 1 were collected
via in-person structured interviews (2007–2010). All six SAGE countries implemented multi-
stage cluster sampling strategies [44]. Household-level and person-level analysis weights, based
on the selection probability at each stage of sampling along with post stratification corrections,
were applied to produce nationally representative cohorts. Age and sex standardizations based
on WHO’s World Standard Population [45] and the United Nations Statistical Division’s pop-
ulation distributions (http://unstats.un.org/unsd/default.htm)) were carried out to adjust for
between country population age and sex differences. Additional details about SAGE are provid-
ed elsewhere [46].
Sample
The sample in this study included adults aged 50 years and over in the six SAGE countries.
There were 29, 807 observations: 11,648 in China; 4,072 in Ghana; 6,350 in India; 2,004 in Me-
xico; 2,933 in the Russian Federation and 2,805 in South Africa. Weighted samples were 30,146
for the six countries comprising: 11,525 in China; 4,059 in Ghana; 6,329 in India; 2,973 in the
Russian Federation and 2,720 in South Africa.
Variables
Dependent Variables. Three dependent variables–past-month back pain prevalence, back
pain frequency/intensity and disability—were derived from questions in the SAGE individual
questionnaire.
Past-month back pain prevalence (no vs. yes) was identified from responses to the question
“Have you experienced back pain in the last 30 days?”
A score measuring the intensity and frequency of past-month back pain, was conditioned
on responses (yes) to this prevalence question (4008). Frequency was measured using re-
sponses to question 4009 which asked “On how many days did you have this back pain during
the last 30 days?” A pain intensity measure was derived by summing responses to questions
2007 and 2008. In these two questions respondents were asked to use a Likert scale (1 = none,
2 = mild, 3 = moderate, 4 = severe and 5 = extreme) to rate the extent of overall bodily pain
(question 2007) and discomfort (question 2008) experienced in the previous 30 days. The in-
tensity and frequency score was computed as the product of these two measures. This was
converted to a percentile index, with zero and 100 indicating the minimum and maximum
possible scores. The distribution was skewed (skewness = 1.48, median 8.3, mean = 18.2 and
standard deviation = 21.6). Scores were grouped into three categories: low (zero score) vs. mod-
erate (> = 50) vs. high intensity (>50). The proportion of respondents in the categories varied
across countries ranging from: 9% to 22% in the low group, 72% to 82% in the moderate group
and 4% to 12% in the high group. This intensity/frequency variable was used here as the mea-
sure of “back pain intensity”.
Disability was measured using the WHO Disability Assessment Schedule (WHODAS 2.0)
encompassing six domains [47, 48]. Twelve items were included in the scale asking about diffi-
culty in functioning in the past 30 days. This included activities of daily living (such as, stand-
ing, dressing oneself) and instrumental activities of daily living (such as learning a new task,
participating in community activities, household chores). Responses were measured on a Likert
scale, ranging from no difficulty to severe difficulty or cannot perform the activity, and
summed to a composite score which was transformed to a scale of 0–100, with 100 indicating
the most severe disability [48].
Socio-demographic variables. Data on socio-demographic characteristics collected in
SAGE were used to describe the study sample. The variables were sex: male vs. female; age: 50–
59 years vs. 60–69 years vs. 70–79 years vs. 80-plus years; education level: no primary complet-
ed vs. completed primary vs. completed secondary or high school vs. completed university or
college; marital status: never married, vs. married/cohabiting vs. divorced/separated/widowed;
work status: never worked vs. currently working vs. not currently working; wealth quintiles: 1
(poorest) to 5 (richest) and area of residence: urban vs. rural. A random-effects probit model
(previously developed and reported elsewhere) was used to estimate wealth levels based on
asset ownership [49, 50]. This was applied to every household in the SAGE surveys and used to
establish country-specific quintiles of household wealth made available by the WHO. The
quintiles provide an alternative measure of income and assets that is less likely to be biased by
contextual differences than traditional income-based measures.
Health-related variables. Health-related variables are described here. Responses to ques-
tions on the use of alcohol and tobacco were categorised as smoker: not current vs. current;
and alcohol: never drinkers vs. former drinkers vs. current drinkers. Obesity was measured
using waist circumference rather than body mass index (BMI). In making this decision we con-
sidered evidence of the importance of waist circumference as a predictor of health outcomes
[30, 51–53] and the availability of WHO recommendations for waist circumference in men
and women separately [54]. The waist circumference variable (low risk vs. high risk) was de-
rived using WHO recommended cut-offs of > = 102 centimetres for men and > = 88 centi-
metres for women.
Physical activity was measured using the Global Physical Activity Questionnaire (GPAQ)
[55, 56] which collects information on sedentary behaviour and physical activity participation
in work, travel, and recreation. A categorical variable measuring low vs. moderate vs. high
physical activity was included. High physical activity was defined as vigorous-intensity activity
(such as running which increases the heart rate and breathing) on at least three days per week,
or seven or more days per week of any combination of walking, moderate or vigorous intensity
activities. Moderate physical activity was defined as (per week) three or more days of vigorous-
intensity activity, or five or more days of moderate-intensity activity (such as walking or
cycling resulting in a small but noticeable increase in the heart rate and breathing) of at least
30 minutes per day, or five or more days of any combination of walking, moderate or vigorous
intensity activities. Low physical activity was defined as not meeting any of these criteria.
Indicator variables (no vs. yes) for symptom-based arthritis, depression, asthma [57, 58]
and self-reported diabetes mellitus are included. Symptom-based conditions were derived
using validated WHO algorithms. A “chronic count” variable (zero vs. one vs. two or more
chronic conditions) was included.
Statistical analyses
Data presented here are weighted and include post stratification adjustments in national coun-
try samples and in the pooled multi-country data set [45].
We undertook a complete case analysis. Data were missing for: education (2.2%); marital
status (0.6%); work status (1.4%); wealth (0.4%); smoking (1.9%): drinking (2.1%); waist cir-
cumference (7.0%); physical activity (1.6%); depression (2.1%); arthritis (2.2%); asthma (2.2%);
diabetes (1.8%) and the count of chronic conditions (1.7%).
Descriptive statistics for socio-demographic variables are presented as proportions for each
country and pooled. The prevalence of back pain and back pain intensity (conditioned on prev-
alence) is shown by countries and pooled, and also by socio-demographic and health-related
characteristics in the pooled sample.
Three sets of multivariable regression were undertaken using the pooled multi-country data
set. Multivariable logistic regression describes association between socio-demographic and
health-related determinants and back pain prevalence. Multivariable ordinal logistic regression
describes association between socio-demographic and health-related determinants and back
pain intensity (low vs. moderate vs. high). Multivariable linear regression describes association
between back pain intensity (independent variable) and disability (dependent variable) adjust-
ing for confounding by socio-demographic and health-related factors. A country variable (ref-
erence China) was included in all multivariable regressions.
The literature includes a substantial number of factors associated with back pain and dis-
ability. In aiming to achieve relatively parsimonious models, we focused on recurrent, com-
monly cited factors, identifiable in the SAGE data. These factors were tested in bivariate
analyses, and where statistically significant (p<0.05), were included as covariates in the
multivariable regressions.
Odds ratios and 95% confidence intervals are reported. Variables were tested for correlation
and multicollinearity. Diagnostic checks were undertaken on models and no violations of as-
sumptions were found. STATA Version 11 (StataCorp, 2009) was used for all statistical
analyses.
Results
Back pain prevalence was estimated on the weighted sample of 30,146 derived from 29,807 in-
dividual observations in the six SAGE countries pooled. In order to measure back pain intensi-
ty, the analysis was conditioned on back pain prevalence, giving a sub-sample of 8,815.
Table 1 compares socio-demographic characteristics of the study sample by countries and
pooled (n = 30,146). There were more females in the study population except in Ghana and
India. In most countries, almost half of the study respondents were aged between 50–59 years.
In contrast to Ghana and India, where over 60% of the respondents reported no primary edu-
cation, almost 18% of respondents in the Russian Federation reported that they had completed
university or college education. The majority of the respondents were married or cohabiting at
the time of the survey. Almost 40% of the respondents were separated, divorced or widowed in
Ghana and the Russian Federation. Over 56% reported that they were currently working in
the Russian Federation compared with 23% in Mexico. About 72% of respondents in India
lived in rural areas, compared with only 22% and 30% in Mexico and the Russian Federation
respectively.
Table 2 presents past-month prevalence (n = 30,146) and back pain intensity conditioned
on prevalence (n = 8,815) for respondents 50 years and older in the six SAGE countries. Over-
all, the self-reported prevalence of back pain in the past month was 30%. Prevalence was high-
est in the Russian Federation (56%) and lowest in China (22%).
Comparing the proportion of respondents in each intensity group, India had the highest
proportion of respondents in the high intensity group (12%) and China and South Africa had
the lowest (4%). In the pooled analysis, 8% of respondents were in the high intensity group,
Table 1. Weighted proportional distribution of socio-demographic characteristics, adults aged 50-plus years, by country and pooled countries,
SAGE Wave 1.
doi:10.1371/journal.pone.0127880.t001
compared with 77% and 15% in the moderate and low intensity groups respectively. Mexico
had the highest prevalence in the low intensity group (22%) and Ghana and India had the low-
est (9%).
Table 3 shows the pooled prevalence of back pain in the past month by socio-demographic
and health-related characteristics (n = 30,146). Past-month back pain prevalence was high for
females (35%), rural dwellers (32%) and those with high risk waist circumference (36%). Preva-
lence was 52% for respondents with arthritis, 55% for depression, 52% for asthma and 31% for
Table 2. Back pain prevalence and intensity, adults aged 50-plus years, by country and pooled, SAGE Wave 1.
doi:10.1371/journal.pone.0127880.t002
diabetes. Amongst respondents with two or more chronic conditions, the prevalence of back
pain was 59%. There are prevalence gradients for wealth, education and age, with higher wealth
and higher education associated with lower prevalence, and older age associated with
higher prevalence.
In Table 3, the proportion of respondents with low intensity (n = 1,311) vs. moderate inten-
sity (n = 6,804) vs. high intensity (n = 700) back pain in the pooled sample is shown by socio-
demographic and health characteristics. There were more females than males were in the high
intensity group (9% vs. 6%). About 10 to 13% of respondents who did not complete primary
education, were separated, divorced or widowed, not working, operated at low levels of physical
activity, and had diabetes, were in the high intensity group. Over 20% of respondents with de-
pression, 18% with asthma, and 17% with two or more chronic conditions were in the high in-
tensity group. An age gradient is evident for the high intensity group; 21% of respondents aged
80 and over had high intensity back pain compared with 5% of respondents aged 50 to
59 years.
Tables 4, 5 and 6 present the results of the adjusted multivariable regressions. Reference cat-
egories for the education, employment, wealth status and physical activity variables were
changed from those shown in Tables 1 and 3 in order to show odds as risk.
In Table 4, female sex, rural residence, being married or separated/divorced/widowed com-
pared with never being married, being a current smoker, and drinking alcohol, were statistically
significantly in association with back pain. There is a gradient in the association between edu-
cation level and back pain. Respondents who had not completed even primary education had
almost two-fold higher odds of reporting back pain compared with those who completed uni-
versity/college. The odds of back pain were 40% higher for respondents in the lowest (poorest
socioeconomic quintile) compared with respondents in the highest quintile (as the reference
group). The odds of back pain increased with the number of chronic comorbidities, (i.e. from
2.7 times for those with one condition, to 4.8 times for those with two or more comorbidities).
There was no statistically significant association between back pain and age, physical activity
level, waist circumference and employment status. Country odds ratios, with China as the ref-
erence, were statistically significant. Adults in the Russian Federation, for example, had four-
fold higher odds of reporting back pain compared with adults in China.
In Table 5, the ordinal logistic regression presents the odds of reporting high intensity back
pain vs. low intensity back pain and high intensity back vs. moderate intensity back pain for
Table 3. Back pain prevalence and intensity by socio-demographic and health characteristics, adults aged 50-plus years, pooled countries SAGE
Wave 1.
Table 3. (Continued)
doi:10.1371/journal.pone.0127880.t003
each of the covariates. Older age, female sex, living in a rural area, not completing primary edu-
cation, not currently working, being a current smoker, and having multiple chronic conditions,
were statistically significant in association with higher back pain intensity. A clear gradient is
seen across age, with individuals aged 80 and over having three times higher odds of high back
pain intensity compared with those aged 50 to 59 years. Unlike the results in Table 4, marital
status and wealth were not statistically significant in association with high back pain intensity.
In Table 6, the multivariable linear regression shows association between levels of back pain
intensity and the continuous WHO disability score. The reference group comprises respon-
dents with no back pain, according to self-reported past-month back pain prevalence. Com-
pared with the non-prevalent reference group, people in the high intensity group had, on
average, a 19-unit worse disability score, and people in the lowest intensity group had less dis-
ability, when all other variables were held constant. The disability score for respondents in the
moderate disability group was, on average, five units higher than the score for those without re-
ported back pain after adjusting for all other variables.
The sample size (N = 29,996) is due to missing data (n = 150) on the intensity score for re-
spondents who were included in prevalence estimates.
Disability was also associated with socioeconomic factors and comorbidities when back
pain intensity was held constant. There are inverse associations between education, wealth and
disability and there was positive association between disability and comorbid chronic condi-
tions. There was higher disability amongst the oldest age group and rural residents. People
with higher physical activity had less disability compared with those with low physical activity,
and those who never worked had higher disability compared to those who were working.
Table 4. Multivariable logistic regression of factors associated with back pain prevalence, adults 50+
years, pooled, SAGE Wave 1 (N = 30,146).
doi:10.1371/journal.pone.0127880.t004
Compared with China, adults aged 50 and older in the other five SAGE countries, had more
disability.
Discussion
This study of the six SAGE countries is the first to utilize nationally representative, comparable,
population survey data to measure and assess factors associated with past-month back pain
prevalence across six culturally different LMICs. These findings are a start but not sufficient.
They serve as a reference point for clinicians, public health practitioners and researchers plan-
ning future qualitative and quantitative studies that can inform development of country-specif-
ic medical education and practice guidelines.
In recent years, there has been increasing recognition of the growing burden of musculo-
skeletal disease and back pain in both developed and developing countries [5, 8, 9, 11, 43]. A
large multi-country study of chronic pain conditions [42] showed that the age standardized
prevalence of chronic pain conditions in the previous twelve months was 37.3% in developed
compared with 41.1% in developing countries, with back pain more common in developing
countries.
Across the six SAGE countries, past-month back pain prevalence was almost 30%. These re-
sults are within range of prevalence estimates reported in some other studies. Estimates of one-
month back pain prevalence for adults aged 60 and over in developing countries range between
18% and 29% (13). In our study prevalence estimates varied across the SAGE countries, from
22% in China to 56% in the Russian Federation. The estimates for the two African countries,
Ghana and South Africa, were 41% and 39% respectively. A review of back pain prevalence
studies in adults aged 20 to 85 years in the African continent reported one-year back pain prev-
alence between 40% and 72% [25]. The high back pain prevalence seen here in the Russian Fed-
eration is consistent with a previous analysis of World Health Survey data in which the
prevalence was 76.8% in the major metropolitan areas of Moscow and St Petersburg [59].
In addition to pharmacological interventions, treatment and management modalities for
back pain include behaviour and exercise therapy and lifestyle change, many of which are rela-
tively low-cost to implement in primary care settings [60]. However, most of what is known
and demonstrated comes from developed countries. Context-specific trials and evaluations in
developing countries are needed.
In agreement with other studies [15, 16, 61] this study shows that back pain increases with
age although not necessarily for the very old, and that female sex is significantly associated
with back pain [16, 18, 24]. The reasons for this are not clear, although it is suggested that this
may be due to greater sensitization to pain, the reporting of pain, and differences in response to
analgesics in females [42, 62, 63]. Even though the mechanisms that lead to gender differences
in pain are yet to be elucidated, in their literature review, Bartley and Fillingim [63] suggested
that multiple bio-psychosocial mechanisms (e.g. genetic, sex hormones, pain coping, gender
roles) may interact and contribute to the phenomenon.
Other studies have also reported inverse socioeconomic gradients between the prevalence
and intensity of back pain and education and wealth [14, 32, 35]. Our study also found that
people living in rural areas were more likely to experience back pain and at higher intensity.
This may have also been due to more frequent and strenuous outdoor household activities (e.g.
carrying water or food), undertaken by older people living in rural areas in these six countries
[14, 36, 37]. The occupational variable in SAGE that identified physical labour however had
considerable missing data, and for that reason, was not included here. Instead, we used a mea-
sure of working status for which the data were over 98% complete. However, a simple cross
Table 5. Multivariable ordinal logistic regression of factors associated with back pain intensity, adults
50+ years, pooled, SAGE Wave 1 (N = 8,815).
doi:10.1371/journal.pone.0127880.t005
tabulation in the pooled SAGE dataset showed that over 51% of rural residents were in occupa-
tions involving physical labour compared to 38% of urban residents.
There was a higher prevalence of back pain among respondents with high risk waist circum-
ferences, although in the presence of other factors in the multivariable models, waist circumfer-
ence was not statistically significant. The precision of these estimates is influenced by missing
data in the waist circumference variable.
In this study of the six SAGE countries there was a higher prevalence of back pain amongst
those with high levels of physical activity and the odds of reporting back pain were slightly
higher for those who had high, compared with moderate or low levels of physical activity.
However, these results were not statistically significant and the association between back pain
and physical activity can occur in both directions. For example Kim et al. [64] found that vigor-
ous and moderate physical activity in older Koreans was associated with an increased risk of
back pain in both men and women, whereas strength exercises were associated with a reduced
risk of back pain.
In the pooled analysis of adults aged 50-plus in the SAGE countries, about 8% of those with
back pain experienced it at high intensity, although at the country level this ranged from 4% in
China to 12% in India. Respondents who experienced high intensity back pain had consider-
ably greater disability, compared with those with low intensity or no back pain. Musculoskele-
tal disorders are a frequent cause of disability in older populations [40] and these findings
reflect the major disabling effect of back pain on daily function and activities. In the multivari-
able disability model, the effect of female sex was not evident as in the other models, but there
was a clear age gradient, with older age significantly associated with greater disability. These re-
sults are generally consistent with other studies. In a study of Korean adults with a mean age of
40 years, Kim et al. [64] found that the degree of disability from back pain assessed was influ-
enced by a pain severity and type. In the US Weiner et al. [65] showed that, in a large cohort of
well-functioning adults aged 70–79 years, back pain frequency/intensity was associated with
perceived difficulty in performing important physical functional tasks. These authors suggested
that the dose-response relationship between back pain frequency/intensity and self-reported
functional task difficulties underscores the importance of efforts to treat and reduce pain with-
out necessarily eradicating it [65].
The WHO measure of disability [48] takes into account variations in the reporting of dis-
ability across cultures [66]. There is increasing evidence of the effectiveness of low-cost easily
implemented therapeutic interventions, such as physical exercise, in improving rehabilitation
outcomes for people with back pain as well as the quality of life for people with disabilities [67]
in higher income countries. The finding of strong association between back pain and disability
in the SAGE countries has important public health policy implications for LMICs. One of the
reasons for this is that the data were population-based being captured in households, rather
than in clinical settings or the workplace. This suggests the need for investment in community-
based primary care assessment and education.
Limitations
The cross sectional nature of the study presents limitations in terms of interpreting causal asso-
ciation. We cannot separate antecedent factors that influence incident cases from consequent
factors associated with prevalent cases. Some determinants may also be consequences, e.g.
smoking and drinking, and there may also be selection effects, e.g. those with back pain are
unable to undertake physical activity at high levels but physical activity may also be a causal
factor. Data from future waves of SAGE will provide information about the direction of
associations.
Table 6. Multivariable regression of factors associated with disability, adults 50+ years, pooled coun-
tries, SAGE Wave 1 (N = 29,996).
Table 6. (Continued)
doi:10.1371/journal.pone.0127880.t006
Although the GPAQ was the best available measure of physical activities in this dataset, it is
possible that we underestimated the amount of physical activities undertaken by these older
adults. The GPAQ only captures work and recreational activities and does not include indoor
and outdoor household activity, which can be significant in developing countries, particularly
in rural areas.
The WHODAS 2.0 disability score captures respondents that have difficulties in performing
everyday tasks that may not be attributable to back pain. Nevertheless, the WHODAS 2.0 was
the most appropriate measure of disability for this study.
We developed an index for back pain intensity and frequency using questions that referred
to pain in general, rather than back pain specifically. While it is true that the pain that the indi-
vidual reports could have been due to a number of conditions, we assume that back pain was
one of these. Therefore if back pain was experienced in the previous 30 days, we assume that it
would have contributed to the responses to the questions about general pain that were used to
develop the intensity/frequency index.
A large proportion of the data in the SAGE were self-reported, and could be influenced by
the reporting heterogeneity by the respondents, either due to their experiences or expectations.
Salomon et al. [68] suggested the use of anchoring vignettes based on fixed levels of health on
different dimensions such as mobility, pain, cognition, to adjust for this. However Hirve et al.
[69] analysed eight health and demographic surveillance sites within the WHO-INDEPTH
Network and found that the use of vignettes to adjust for reporting heterogeneity could not be
justified because vignette equivalence and response consistency requirements were
not fulfilled.
Because there was a relatively low percentage of missing data, we decided against using mul-
tiple imputation methods. We acknowledge the possibility of bias due to missing data.
Strengths
Measures of back pain prevalence are typically based on self-report. Estimates vary widely
across populations and settings owing to methodological, definitional and socio-cultural differ-
ences [11, 12, 36]. Socio-cultural and psychosocial factors influence the reporting of pain, as
well as knowledge and perception of ways of dealing with the impact of pain on everyday func-
tioning [1, 41]. The SAGE adjusted for cultural differences as far as possible by using standard-
ized culturally appropriate instruments.
This is the first study of its kind to use nationally representative standardized population
survey data to present detailed contextual analyses of back pain and disability in older adults in
LMICs. The questionnaire was first translated into the local language, and then back translated.
All translations were validated before data collection commenced. The interviews were
Conclusions
Our study highlights the need to further gather data and investigate back pain determinants in
older adults within country settings [5]. We do not yet fully understand the impact of sociocul-
tural factors on the perception and reporting of back pain in developing countries [29]. Our
findings are a start but certainly not sufficient. What is needed is context-specific data that can
inform the design, development, trialling and economic evaluation of interventions for the di-
agnosis, management, treatment and rehabilitation of back pain in older adults in developing
countries.
Acknowledgments
We thank the respondents in each country for their continued contributions, and acknowledge
the expertise and contributions of the country principal investigators and their respective
survey teams. The US National Institute on Aging’s Division of Behavioral and Social
Research, under the directorship of Dr Richard Suzman, has been instrumental in providing
continuous support to SAGE and has made the entire endeavour possible. We thank our SAGE
Collaborators. Their names are listed as follows: Ghana: Mensah GP, Department of Commu-
nity Health, University of Ghana Medical School. China: China; Guo Y and Zheng Y, Shanghai
Municipal Center for Disease Control and Prevention, Shanghai, China. Yong J, National Cen-
ter for Chronic and Noncommunicable Disease Control and Prevention, China Center for Dis-
ease Control and Prevention, Beijing. India: Parasuraman P, Lhungdim H and Sekher TV,
International Institute for Population Sciences, Mumbai, India. Mexico: Rosa R, Lopez Ridaura
R, Salinas-Rodriguez A, and Manrique-Espinoza B, National Institute of Public Health, Cuer-
navaca, Mexico. Russian Federation: Belov VB, Lushkina NP, and Meshkov D, National Re-
search Institute of Public Health (FSBI, RAMS) Moscow, Russian Federation. South Africa:
Phaswana-Mafuya R, Makiwane M, Zuma K, Ramlagan S, Davids A, Mbelle N and Matseke G,
Human Sciences Research Council, Pretoria, South Africa; Schneider M, independent consul-
tant; Tabane C, University of the Witwatersrand, Johannesburg, South Africa. USA: Snodgrass
JJ, Eick G, Sterner K, Liebert MA, Gilbert TE, University of Oregon, University of Oregon, De-
partment of Anthropology, Eugene, Oregon, USA: Thiele EA, Department of Biology, Vassar
College, Poughkeepsie, New York USA. Australia: Byles JE, Research Centre for Gender,
Health and Ageing; University of Newcastle, Newcastle, Australia. D'Este C, National Centre
for Epidemiology & Population Health, Australian National University, Canberra, Australia.
Negin J. School of Public Health, University of Sydney, Sydney, Australia.
In this work, Jennifer Stewart Williams and Nawi Ng were supported by the FORTE grant
for the Umeå Centre for Global Health Research (No. 2006–1512). Nawi Ng was also sup-
ported by the Swedish Research Council’s “Linnestöd” grant (No 2006-21576-35119-66) for
the Ageing and Living Conditions Programme at Umeå University, Sweden. The funders had
no role in study design, data collection and analysis, decision to publish, or preparation of
the manuscript.
Author Contributions
Conceived and designed the experiments: JSW NN SC. Analyzed the data: JSW NN SC. Wrote
the paper: JSW NN SC. Contributed to the design and implementation of the study, data collec-
tion and review of the manuscript: KP AY RB TM FW PA PK.
References
1. Woolf AD, Pfleger B. Burden of major musculoskeletal conditions. Bull World Health Organization.
2003; 81(9):646–56. PMID: 14710506
2. Waddell G. Volvo award in clinical sciences. A new clinical model for the treatment of low-back pain.
Spine (Phila Pa 1976). 1987; 12:632–44. PMID: 2961080
3. Borrell-Carrió F, Suchman AL, Epstein RM. The Biopsychosocial Model 25 years later: principles, prac-
tice, and scientific Inquiry. Ann Fam Medicine. 2004; 2:576–82. PMID: 15576544
4. Gureje O, Von Korff M, Simon GE, Gater R. Persistent pain and well-being. A World Health Organiza-
tion Study in Primary Care. JAMA. 1998; 280(2):147–51. PMID: 9669787
5. Brooks P. Issues with chronic musculoskeletal pain. Rheumatology. 2005; 44:831–3. PMID: 15840598
6. Rudy TE, Weiner DK, Lieber SJ, Slaboda J, Boston JR. The impact of chronic low back pain on older
adults: A comparative study of patients and controls. Pain. 2007; 131(3):293–301. PMID: 17317008
7. Weiner SS, Nordin M. Prevention and management of chronic back pain. Best Practice & Research
Clinical Rheumatology. 2010; 24:267–79.
8. Hoy D, March L, Brooks P, Blyth F, Woolf A, Bain C, et al. The global burden of low back pain: estimates
from the Global Burden of Disease 2010 study. Ann Rheum Dis. 2014. Epub 24 March 2014.
9. Murray CJL, Vos T, Lozano R, Naghavi M, Flaxman AD, Michaud C, et al. Disability-adjusted life years
(DALYs) for 291 diseases and injuries in 21 regions, 1990–2010: a systematic analysis for the Global
Burden of Disease Study 2010. The Lancet. 2012; 380:2198–227.
10. Institute for Health Metrics and Evaluation. http://www.healthmetricsandevaluation.org/gbd Seattle: Bill
and Melinda Gates Foundation; 2013 [6 June 2013].
11. Brooks PM. The burden of musculoskeletal disease—a global perspective. Clinical Rheumatology.
2006; 25 778–81. PMID: 16609823
12. Manek NJ, MacGregor AJ. Epidemiology of back disorders: prevalence, risk factors, and prognosis.
Curr Opin Rheumatol. 2005; 17:134–40. PMID: 15711224
13. Dionne CE, Dunn KM, Croft PR, Nachemson AL, Buchbinder R, Walker BF, et al. A consensus ap-
proach toward the standardization of back pain definitions for use in prevalence studies. Spine. 2008;
33(1):95–103. doi: 10.1097/BRS.0b013e31815e7f94 PMID: 18165754
14. Hoy D, Brooks P, Blyth F, Buchbinder l. The Epidemiology of low back pain. Best Pract R Res Clin
Rheumatol. 2010; 24(6):769–81. doi: 10.1016/j.berh.2010.10.002 PMID: 21665125
15. Knauer SR, Freburger JK, Carey TS. Chronic low back pain among older adults: a population-based
perspective. Journal of Ageing and Health. 2010; 22(8):1213–34. doi: 10.1177/0898264310374111
PMID: 20657007
16. Cecchi F, Debolini P, Molino Lova R, Macchi C, Bandinelli S, Bartali B, et al. Epidemiology of back pain
in a representative cohort of Italian persons 65 years of age and older: The InCHIANTI Study. Spine.
2006; 31(10):1149–55. PMID: 16648752
17. Fejer R, Ruhe A. What is the prevalence of musculoskeletal problems in the elderly population in devel-
oped countries? A systematic critical literature review. Chiropractic & Manual Therapies. 2012; 20
(31):52.
18. Jacobs JM, Hammerman-Rozenberg R, Cohen A, Stessman J. Chronic Back Pain Among the Elderly:
Prevalence, Associations, and Predictors. Spine. 2006; 31(7):E2003–E7.
19. Dionne CE, Dunne KM, Crift PR. Does back pain prevalence really decrease with increasing age? A
systematic review. Age and Ageing. 2006; 35(3):229–34.
20. Dillon C, Paulose-Ram R, Hirsch R, Gu QS. Skeletal muscle relaxant use in the United States: data
from the Third National Health and Nutrition Examination Survey (NHANES III). Spine. 2004; 29.
(8):892–6. PMID: 15082991
21. Moore JE. Chronic low back pain and psychosocial issues. Phys Med Rehabil Clin N Am 2010;
21:801–15. doi: 10.1016/j.pmr.2010.06.005 PMID: 20977962
22. Stranjalis G, Tsamandouraki K, Sakas DE, Alamanos Y. Low back pain in a representative sample of
Greek population. Spine. 2004; 29(12):1355–61. PMID: 15187638
23. Raspe H, Matthis C, Croft P, O’Neill T, European Vertebral Osteoporosis Study Group. Variation in
back pain between countries. Spine. 2004; 29(9):1017–21. PMID: 15105675
24. Donald IP, Foy C. A longitudinal study of joint pain in older people. Rheumatology. 2004; 43:1256–60.
PMID: 15252209
25. Louw. The prevalence of low back pain in Africa: a systematic review. BMC Musculoskeletal Disorders.
2007; 8(105):14. PMID: 17976240
26. Han TS, Schouten JS, Lean ME, Seidell JC. The prevalence of low back pain and associations with
body fatness, fat distribution and height. Int J Obes Relat Metab Disord. 1997; 21:600–7. PMID:
9226492
27. Goldberg MS, Scott SC, Mayo NE. A review of the association between cigarette smoking and the de-
velopment of nonspecific back pain and related outcomes. Spine. 2000; 25(8):995–1014. PMID:
10767814
28. Deltonen M, Lindross AK, Torgerson JS. Musculoskeletal pain in the obese: a comparison with a gener-
al population and long term changes after conventional and surgical obesity treatment. Pain. 2003;
104:549–57. PMID: 12927627
29. Gilgil E, Kaçar C, Bütün B, Tuncer T, Urhan S, Yildirim C, et al. Prevalence of low back pain in a devel-
oping urban setting. Spine. 2005; 30(9):1093–8. PMID: 15864165
30. Shiri R, Karppinen J, Leino-Arjas P, Solovieva S, Viikari-Juntura E. The association between obesity
and low back pain: a meta-analysis. American Journal of Epidemiology. 2010; 171(2):135–54. doi: 10.
1093/aje/kwp356 PMID: 20007994
31. Muraki S, Akune T, Oka H, En-Yo Y, Yoshida M, Ishibashi H, et al. Health-related quality of life in sub-
jects with low back pain and knee pain in a population-based cohort study of Japanese men: the Re-
search on Osteoarthritis Against Disability study. Spine. 2011; 36(16):1312–9. doi: 10.1097/BRS.
0b013e3181fa60d1 PMID: 21730819
32. Heistaro S, Vartiainen E, Heliövaara, Puska P. Trends of back pain in Eastern Finland, 1972–1992, in
relation to socioeconomic status and behavioral risk factors. American Journal of Epidemiology. 1998;
148(7):671–82. PMID: 9778174
33. Plouvier S, Leclerc A, Chastang J-F, Bonefant S, Goldberg M. Socioeconomic position and low back
pain—the role of biomechanical strains and psychosocial work factors in ther GAZEL cohort. Scan J
Work Environ Health. 1999; 35(6):429–36. PMID: 19806277
34. Latza U, Kohlmann T, Deck R, Heiner Raspe H. Can health care utilization explain the association be-
tween socioeconomic status and back pain? Spine. 2004; 29(14):1561–6. PMID: 15247579
35. Carr JL, Klaber Moffett JA. Review Paper: The impact of social deprivation on chronic back pain out-
comes. Chronic Illness. 2005; 1:121–9. PMID: 17136918
36. Cho NH, Jung YO, Lim SH, Chung C-K, Kim HA. The prevalence and risk factors of low back pain in
rural community residents of Korea. Spine. 2012; 37(24):2001–10. doi: 10.1097/BRS.
0b013e31825d1fa8 PMID: 22588379
37. Guo H-R, Chang Y-C, Yeh W-Y, Chen C-W, Guo YL. Prevelance of musculoskeletal disorder among
workers in Taiwan: a nationwide study. Journal of Occupational Health. 2004; 46:26–36. PMID:
14960827
38. Jin K, Sorock GS, Courtney TK. Prevalence of low back pain in three occupational groups in Shanghai,
People’s Republic of China. Journal of Safety Research. 2004; 35:23–8. PMID: 14992843
39. Karunanayake AL, Pathmeswaran A, Kasturiratne A, Wijeyaratne LS. Risk factors for chronic low back
pain in a sample of suburban Sri Lankan adult males. International Journal of Rheum Dis. 2013; 16
(2):203–10. doi: 10.1111/1756-185X.12060 PMID: 23773646
40. World Health Organization. The burden of musculoskeletal conditions at the start of the new millenni-
um. Geneva: 2003.
41. Volinn E. The epidemiology of low back pain in the rest of the world: a review of surveys in low-and mid-
dle-income countries. Spine. 1997; 22(15):1747–54. PMID: 9259786
42. Tsang A, Von Korff M, Lee S, Alonso J, Karam E, Angermeyer MC, et al. Common chronic pain condi-
tions in developed and developing countries: gender and age differences and comorbidity with depres-
sion-anxiety disorders. The Journal of Pain. 2008; 9(10):883–91. doi: 10.1016/j.jpain.2008.05.005
PMID: 18602869
43. Hoy D, March L, Brooks P, Woolf A, Blyth F, Vos T, et al. Measuring the global burden of low back pain.
Best Practice & Research Clinical Rheumatology. 2010; 24:155–65.
44. Naidoo N. Working Paper Number 5. Genève: WHO SAGE, 2012.
45. Ahmad OB, Boschi-Pinto C, Lopez AD, Murray CJL, Lozano R, Inoue M. Age standardization of rates:
a new WHO standard. Geneva: World Health Organization, 2001.
46. Kowal P, Chatterji S, Naidoo N, Biritwum R, Wu F. Data resource profile: The World Health Organiza-
tion Study on global AGEing and adult health (SAGE). Int J Epidemiol. 2012; 41:1639–49. doi: 10.1093/
ije/dys210 PMID: 23283715
47. Ütsun TB, Chatterji S, Bickenbach J, J K, Schneider M. The International Classification of Functioning,
Disability and Health: a new tool for understanding disability and health. Disability and Rehabilitation.
2003; 25(11–12):656–571. PMID: 14617445
48. Ütsun TB, Kostanjsek N, Chatterji S, Rehm J. Measuring health and disability: manual for WHO Disabil-
ity Assessment Schedule (WHODAS 2.0). Geneva, Switzerland: World Health Organization, 2010.
49. Ferguson B, Murray CL, Tandon A, Gakidou E. Estimating permanent income using asset and indicator
variables. In: Murray CL, Evans DB, editors. Health systems performance assessment debates, meth-
ods and empiricism. Geneva: World Health Organization; 2003.
50. Howe LD, Galobardes B, Matijasevich A, Gordon D, Johnston D, Onwujekwe O, et al. Measuring socio-
economic position for epidemiological studies in low- and middle-income countries: a methods of mea-
surement in epidemiology paper. International Journal of Epidemiology. 2012; doi: 10.1093/ije/
dys037:16
51. Lee CMY, Huxley RR, Wildman RP, Woodward M. Indices of abdominal obesity are better discrimina-
tors of cardiovascular risk factors than BMI: a meta-analysis. Journal of Clinical Epidemiology. 2008;
61:646–53. doi: 10.1016/j.jclinepi.2007.08.012 PMID: 18359190
52. Huxley R, Mendis S, Zheleznyakov E, Reddy S, Chan J. Body mass index, waist circumference and
waist:hip ratio as predictors of cardiovascukar risk-a review of the literature. European Journal of Clini-
cal Nutrition. 2010; 64:16–22. doi: 10.1038/ejcn.2009.68 PMID: 19654593
53. Sherf Dagan S, Segev S, Novikov I, Dankner R. Waist circumference vs body mass index in associai-
ton with cardiorespiratory fitness in healthy men and women: a cross sectional analysis of 403 subjects.
Nutritional Journal. 2013; 12(12):8.
54. World Health Organization. Waist circumference and waist-hip ratio. Genève: WHO, 2008 8–11 De-
cember. Report No.
55. World Health Organization. Global Physical Activity Questionnaire (GPAQ). Analysis Guide. Surveil-
lance and Population-Based Prevention. Genève: Department of Chronic Diseases and Health Pro-
motion, WHO.
56. World Health Organization. Global strategy on diet, physical activity and health. Geneva: WHO, 2004
22 May 2004. Report No.
57. Moussavi S, Chatterji S, Verdes E, Tandon A, Patel V, Ustun B. Depression, chronic diseases, and dec-
rements in health: results from the World Health Surveys. The Lancet. 2007; 370:851–8.
58. He W, Muenchrath M, Kowal P. Shades of Gray: A Cross-Country Study of Health and Well-Being of
the Older Populations in SAGE Countries, 2007–2010. Washington DC: US Department of Health and
Human Services, 2012 Contract No.: P95/12-01.
59. Maximova TM, Belov VB, Lushkina NP, Barabanova NA. Health and medical care problems of people
60 years and over. I European Congress Advances in Gerontology. 2007; 20(3):273.
60. Connelly LB, Woolf A, Brooks P. Cost-effectiveness of interventions for musculoskeletal conditions. In:
Jamison DT BJ, Measham AR, editor. Disease Control Priorities in Developing Countries. 2nd ed.
Washington (DC): World Bank; 2006.
61. Scheele J, Luijsterburg PAJ, Bierma-Zeinstra SMA, Koes BW. Chronic low back pain among older
adults: A population-based perspective. Eur J Phys Med Rehab. 2012; 48:379–86.
62. Breivik H, Collett B, Ventafridda V, Cohen R, Gallacher D. Survey of chronic pain in Europe: Preva-
lence, impact on daily life, and treatment. Eur J Pain. 2006; 10:287–333. PMID: 16095934
63. Bartley EJ, Fillingim RB. Sex differences in pain: a brief review of clinical and experimental findings.
British Journal of Anaesthesia. 2013; 111(1):52–8. doi: 10.1093/bja/aet127 PMID: 23794645
64. Kim GS, Yi C, Cynn H. Factors influencing disability due to low back pain using the Oswestry Disability
Questionnaire and the Quebec Back Pain Disability Scale. Physiotherapy Research International.
2014. Epub March.
65. Weiner DK, Haggerty CL, Kritchevsky SB, Harris T, Simonsick EM, Nevitt M, et al. How does low back
pain impact physical function in independent, well-functioning older adults? Evidence from the Health
ABC Cohort and Implications for the future. Pain Medicine. 2003; 4(4):311–20. PMID: 14750907
66. Sousa RM, Dewey ME, Acosta D, Jotheeswaran AT, Castro-Costa E, Ferri CP, et al. Measuring disabil-
ity across cultures—the psychometric properties of the WHODAS II in older people from seven low and
middle income countries. The 10/66 Dementia Research Group population based survey. Int J Methods
Psychiatr Res. 2010; 19(1):1–17. doi: 10.1002/mpr.299 PMID: 20104493
67. World Health Organization, The World Bank. World report on disability. Geneva: WHO 2011.
68. Salomon JA, Tandon A, Murray CJL, for the World Health Survey Pilot Study Collaborating Group.
Comparability of self rated health: cross sectional multi-country survey using anchoring vignettes. BMJ.
2004; 328(7434):258–61. PMID: 14742348
69. Hirve S, Gómez O, X, Oti S, Debpuur C, Juvekar S, Tollman SM, et al. Use of anchoring vignettes to
evaluate health reporting behavior amongst adults aged 50 years and above in Africa and Asia testing
assumptions. Global Health Action. 2013.