- Email: [email protected]
Ten year course of low back pain in an adult population-based cohort – The Doetinchem Cohort Study
European Journal of Pain 15 (2011) 993–998
Contents lists available at ScienceDirect
European Journal of Pain journal homepage: www.EuropeanJournalPain.com
Ten year course of low back pain in an adult population-based cohort – The Doetinchem Cohort Study Sandra H. van Oostrom a,⇑, W.M. Monique Verschuren a, Henrica C.W. de Vet b, H. Susan J. Picavet a a b
National Institute of Public Health and the Environment, Bilthoven, The Netherlands Department of Epidemiology and Biostatistics and the EMGO Institute for Health and Care Research, VU University Medical Centre, Amsterdam, The Netherlands
a r t i c l e
i n f o
Article history: Received 2 August 2010 Received in revised form 18 January 2011 Accepted 24 February 2011 Available online 22 March 2011 Keywords: Low back pain Persistence Prospective cohort Prevalence Course
a b s t r a c t Background: The long-term course of long-standing low back pain is largely unknown since long-term data are scarce. Objective: We examined the course of self-reported low back pain in the prospective population-based Doetinchem cohort over a period of 10 years. Methods: Between 1993 and 2007, around 5700 randomly selected men and women in four age groups of originally 20–29, 30–39, 40–49, 50–59 years were measured three times. Logistic regression analysis was used to study the association of sociodemographic (gender, age, education, work status) and lifestyle characteristics (BMI, smoking, physical activity) with persistent and new episodes of long-standing low back pain. Results: The prevalence of long-standing low back pain is quite stable over a 10 year period, approximately 20% on population level. On individual level, around 30% of the population was completely free of low back pain during the entire period, 6% can be characterized as persistent back pain sufferers. Individuals with persistent and a varying pattern have a more unhealthy lifestyle (BMI and smoking) than those without low back pain. Age, smoking, obesity and not having a paid job are associated with 10-year persistent back pain in the general population, whereas age and not having a paid job are associated in those with long-lasting back pain at baseline. New episodes of long-standing back pain are relatively frequent among women and smokers. Conclusions: Low back pain in the population is characterized as very dynamic which challenges epidemiological studies highly. Long-term information on the course of back pain is needed to define severe subgroups. Ó 2011 European Federation of International Association for the Study of Pain Chapters. Published by Elsevier Ltd. All rights reserved.
1. Introduction Low back pain (LBP) is a major health problem in developed countries; more than 70% of the population will experience an episode of LBP at some time in their lives (Rubin, 2007). Estimates of prevalence of LBP over longer periods of time showed an increase (Freburger et al., 2009; Harkness et al., 2005; Leijon and Mulder, 2009), a decrease (Puts et al., 2008; Huppe et al., 2007) or a stable prevalence over time (Leino et al., 1994). The course of LBP is characterized by variation and change (Von Korff and Saunders, 1996). Acute LBP generally resolves within weeks but recurrences and long-term persistence of the pain are ⇑ Corresponding author. Address: Department for Prevention and Health Services Research (PZO, pb 101), National Institute of Public Health and the Environment, PO box 1, 3720 BA Bilthoven, The Netherlands. Tel.: +31 30 2748596; fax: +31 30 2744407. E-mail address: [email protected] (S.H. van Oostrom).
common (Hestbaek et al., 2006b; Dunn et al., 2006). Recent estimates showed an increase of chronic LBP to 10% of the population in the US (Freburger et al., 2009). Individuals with chronic LBP report high pain levels, high health care utilization and frequent work disability (van Tulder et al., 1998; Hagen and Thune, 1998; Picavet et al., 2008). Chronic LBP is therefore referred to as a severe and costly condition. However, the long-term course of chronic LBP is largely unknown. It is suggested that individuals with chronic, long-standing LBP, which is usually defined as LBP for more than 3 months, tend to show a more persistent course (Hayden et al., 2010). However, even in long-standing LBP pain might not persist during the entire life course. Follow-up periods of most prospective studies investigating the course of LBP are limited to 1 year, which can be regarded as relatively short for a dynamic condition as LBP (Dunn et al., 2006). Understanding the long-term course of LBP is important because it provides information on the need for prevention and treatment. Life long prospective studies in the general population are
1090-3801/$36.00 Ó 2011 European Federation of International Association for the Study of Pain Chapters. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ejpain.2011.02.007
994
S.H. van Oostrom et al. / European Journal of Pain 15 (2011) 993–998
regarded the optimal method for studying the course of LBP (Hestbaek et al., 2003). Unfortunately, life long prospective studies are scarce. The ability to predict which individuals are susceptible to long-standing LBP would be a major step ahead with regard to prevention. Recent studies point out that unhealthy lifestyles such as smoking and physical inactivity are associated with LBP (Hestbaek et al., 2006a; Mitchell et al., 2010; Bjorck-van Dijken et al., 2008). Lifestyle-factors might be important elements of prevention, since they are modifiable in nature. Yet, the role of lifestyle factors for long-standing LBP is unclear. The objective of the current study is to investigate the course of self-reported LBP over 10 years in the Doetinchem Cohort Study, a prospective study in the general population in the Netherlands. Sociodemographic and lifestyle characteristics of LBP patterns over time were investigated. Furthermore, we studied whether 10-year persistent LBP and new episodes of long-standing LBP can be predicted based on simple sociodemographic and lifestyle characteristics. 2. Methods 2.1. Study population The Doetinchem Cohort Study is a prospective observational longitudinal population-based study with 15 years of follow-up. First measurements (t0) took place between 1987 and 1991 (see flow chart in Fig. 1). In that period 12,405 inhabitants of Doetinchem between 20 and 60 years old were examined as part of the ‘Monitoring Project on Cardiovascular Disease Risk Factors’ (MP-CVDRF). From the participants of the first examination (t0), a random sample of 7769 was invited to participate for a second examination in the Monitoring Project on Chronic Disease Risk Factors (MORGEN-project) (t1: 1993–1997), of which 79% (n = 6118) participated (Verschuren et al., 2008). From the second examination onwards a measurement of LBP was included. Therefore, the second examination forms the baseline (t1) of our analyses. Those
Round 1, t0 1987 1991 1987-1991
In Doetinch Doetinchem 20,155 20 155 adults 20-59 yrs were invited 12,405 12 405 (62%) invited, were examined
Round 2, t1 1993-1997, baseline
7,769 were invited, 6,118 i d (79%) were examined
Round 3, t2 1998-2002 1
6,579 were invited, 4,917 (75%) were examined
Round 4, t3 2003-2007
5,783 were invited, 4,520 (78%) were examined
invited for the second examination, were invited for the follow-up measurements 5 and 10 years later for a third (t2: 1998–2002) and fourth examination (t3: 2003–2007). Participants who actively refused to participate in the second or third examination were not invited again. All participants who were invited for t3 and participated in t1, t2 or t3 were selected for this study, ending up with a total of 5706 participants. The study was approved by the Medical Ethics Committee of the Netherlands Organization of Applied Scientific Research Institute. All participants gave written informed consent. 2.2. Data collection and processing Measurements included questionnaires and a physical examination. Details on sampling and data collection procedures are described elsewhere (Verschuren et al., 2008). Information about sociodemographic characteristics, lifestyle factors and medical history of (chronic) diseases was obtained by self-report. Sociodemographic and lifestyle factors as reported at baseline were used for this study. Age was categorized in four 10-year categories: 26–35, 36–45, 46–55, and 56–65 years old. Educational level was assessed as the highest level reached and classified into three categories: low (intermediate secondary education or less), medium (intermediate vocational or higher secondary education) and high (higher vocational education or university). Smoking was categorized as current smoker or non-smoker, the latter included ex-smokers. Physical activity was assessed by an extensive physical activity questionnaire, from which the total time spent on moderate-to-vigorous physical activities was calculated. Those who met the recommended level of 3.5 h a week spent on at least moderately intense physical activity were categorized as being active, and those who did not as being inactive. Body mass index (BMI) was calculated from measured height (m) and body weight (kg). Overweight was defined as a BMI between 25 and 30 kg/m2, and obesity was defined as a BMI equal to or above 30 kg/m2.
This study consisted of 5,706 participants who were invited for t3 and participated i t1 in t1, t2 or t3 t3. The complete case anal analysis sis consisted of 4,007 participants who completed all three examinations. examinations
Fig. 1. Flow chart of the Doetinchem Cohort Study.
S.H. van Oostrom et al. / European Journal of Pain 15 (2011) 993–998
2.3. Measurement of low back pain The 12-month period prevalence of LBP and long-standing LBP were measured at t1, t2 and t3. The 12-month prevalence of LBP was measured with the question ‘Have you had trouble, discomfort or pain in the lower part of the back during the last 12 months?’ that was based on the Nordic back questionnaire (Kuorinka et al., 1987). This question was followed by questions regarding LBP at the moment, the number of episodes of LBP in the past year and the duration of LBP. These questions were used to create a measure for long-standing LBP, which is the main focus of this study. The question about the duration of LBP was formulated like ‘In the past 12 months, how long in total did you have LBP?’, with the following seven response options ‘less than a week’, ‘1 or 2 weeks’, ‘3 or 4 weeks’, ‘5 or 6 weeks’, ‘7–12 weeks’, ‘more than 12 weeks’ and ‘pain is always present’. Long-standing LBP was assigned at t1 when there was one episode of LBP lasting more than 12 weeks or when pain is always present during the past year. At t2 the question about the duration of LBP was formulated like ‘How long have you had LBP for at the moment?’ with the following three response options ‘a few days’, ‘less than a month’ and ‘more than a month’. At t3 the question about the duration of LBP was formulated like ‘Has your current LBP lasted for more than 3 months?’. Due to different phrasing at the measurements different cutoff points were used to define whether LBP was long-standing: at t1 and t3 the cutoff point was more than 3 months, at t2 more than 1 month. The 12-month period prevalence of LBP and the prevalence of long-standing LBP at the three rounds were calculated. Five patterns of long-standing LBP were composed based on the three measurements: (1) no long-standing LBP = no long-standing LBP at t1, t2 and t3; (2) new episode of long-standing LBP = no long-standing LBP at t1 and long-standing LBP at t2 or t3; (3) recovery of long-standing LBP = long-standing LBP at t1 and no long-standing LBP at t2 or t3; (4) variable long-standing LBP = the reporting of long-standing LBP varied each subsequent measurement; (5) persistent LBP = long-standing LBP at t1, t2 and t3.
995
the measurement of physical activity changed after this year. All statistical analyses were conducted in SAS software version 9.1.
2.5. Missing data and multiple imputations Age and sex were available for all participants in our study. About 4.6% of the participants did not participate at t1, 15.8% and 20.8% of the participants had complete missing data at t2 and t3. Complete case analysis, including participants who participated in all three measurements, may lead to biased results since data are unlikely to be ‘missing completely at random’(Sterne et al., 2009; Klebanoff and Cole, 2008). Data are ‘missing completely at random’ when no systematic differences exist between the missing values and observed values. When systematic differences exist that can be explained by differences in observed data the data are ‘missing at random’. At t2 and t3, responders have been compared to the non-responders with respect to several variables (Blokstra et al., 2010). Compared to responders, non-responders were lower educated, obese, smokers, and less active. Therefore, we assumed that our data are ‘missing at random’ and used the Multivariate Imputation by Chained Equations (MICE) package in R to perform multiple imputation (van Buuren and Groothuis-Oudshoorn, in press). MICE allows dichotomous, continuous, and categorized variables. In the MICE procedure logistic regression is applied to impute incomplete dichotomous variables, linear regression to impute continuous variables, and predictive mean matching to impute categorized variables. The imputation matrix consisted of all sociodemographic and lifestyle characteristics at baseline, and the LBP variables at baseline, t2 and t3. Five imputation sets were created, analyzed by logistic regression (as described above), and pooled by the MIANALYZE procedure in SAS. The chi-square tests were pooled by the combchi macro (Schafer, 1997; Allison, 2001). Baseline measurements of complete and imputed cases were compared based on means and frequencies. Results from the complete cases and imputed logistic regression analyses were compared and imputed analyses are presented.
3. Results 2.4. Statistical analyses First, the frequency and percentage for each of the LBP patterns were calculated. Second, differences in sociodemographic and lifestyle characteristics were presented and tested by chi-square statistics for (1) no long-standing LBP, (2) persistent LBP over 10 years and (3) varying (episodic) LBP which consisted of a new episode of long-standing LBP, recovery of long-standing LBP and variable long-standing LBP over the measurements. Third, to identify high risk groups for long-standing LBP we determined which sociodemographic and lifestyle characteristics were associated with 10-year persistent and new episodes of long-standing LBP in three logistic regression models: (1) persistent LBP over 10 years in the general adult population; (2) persistent LBP over 10 years among those with long-standing LBP at baseline (t1); (3) new episodes of long-standing LBP in 10 years among those without long-standing LBP at baseline (t1). Sociodemographic characteristics (gender, age, educational level, and work status) and lifestyle factors (body mass category, smoking status, and physical activity) at t1 were included in the logistic regression analyses as potential determinants of persistent or new episodes of LBP. For the logistic regression analyses, 968 respondents who were measured in 1993 of t1 were excluded because
The study population consists of 2686 men and 3020 women aged 26–65 years (mean (SD) 45.9 (10.0)) at baseline. Characteristics of the complete cases (n = 4007) and all participants including imputed cases (n = 5706) are shown in Table 1. Twelve-month period prevalence of LBP was reported by 45% at baseline, while long-standing LBP was reported by 20% respectively (Table 1). On population level, the prevalences of LBP and long-standing LBP remained quite stable over the 10 year period. About 71% ever reported any LBP in three measurements over a period of 10 years, indicating that 29% of the population was free of LBP during the three response periods. Among those reporting any LBP, half of them (37%) reported an episode of long-standing LBP. Table 2 shows the patterns of long-standing LBP over 10 years. About 6% can be characterized as persistent LBP sufferers, 62% had no period of long-standing LBP during the 10 year follow-up, 10% recovered from long-standing LBP, 11% reported a new episode of long-standing LBP, and another 11% had a variable pattern of long-standing LBP. Differences between the patterns no longstanding LBP, persistent LBP and varying LBP were found on all sociodemographic and lifestyle characteristics, except for physical activity. Compared to those without LBP, participants with persistent LBP and a varying pattern of LBP were more often female, older, lower educated, smoker, more frequently overweight and obese, and not having a paid job (Table 3).
996
S.H. van Oostrom et al. / European Journal of Pain 15 (2011) 993–998
Table 1 Sociodemographic characteristics, lifestyle characteristics and LBP at t1 for complete cases and imputed cases. Mean and standard deviation are presented for continuous variables, otherwise percentages are presented. Complete cases (n = 4007) (%)
Imputed cases (n = 5706) (%)
Sociodemographic characteristics Age Men
45.8 (9.8) 47.4
45.9 (10.0) 47.1
Educational level Low Medium High Paid job
51.0 27.4 21.6 63.9
54.7 26.5 18.8 61.3
Lifestyle characteristics Smoking Physical active BMI
27.6 78.9 25.3 (3.6)
31.2 76.7 25.5 (3.7)
Low back pain 12-Month period prevalence (t1) 12-Month period prevalence (t2) 12-Month period prevalence (t3) Long-standing pain (t1) Long-standing pain (t2) Long-standing pain (t3)
45.3 49.0 46.6 17.4 17.4 19.9
45.7 49.4 47.1 18.7 18.4 20.6
LBP patterns over 10 years Ever LBP Ever long-standing LBP Persistent LBP over 10 years New episode of long-standing LBP
70.4 35.7 4.6 18.3
70.6 37.1 5.3 18.3
Table 2 Long-standing LBP patterns of the participants of the Doetinchem Cohort Study over a period of 10 years after multiple imputation. Long-standing LBP
Percentage (n = 5706) (%)
t1
t2
t3
No long-standing LBP New episode of long-standing LBP
No No No
No Yes No
No Yes Yes
62.4 10.8
Recovery of long-standing LBP
Yes Yes
Yes No
No No
10.3
Variable course of long-standing LBP
Yes No
No Yes
Yes No
10.9
Persistent LBP over 10 years
Yes
Yes
Yes
5.6%
Ten-year persistent LBP is predicted by older age, smoking, obesity and not having a paid job at baseline (Table 4, model 1). Among those with long-standing LBP at baseline those with persistent LBP were predicted by being older and smoker, and not having a paid job (model 2). Model 3 investigated the predictors for new episodes of long-standing LBP among participants without longstanding LBP at t1. Being a woman and smoker were predictive for a new episode of long-standing LBP at the 10-year follow-up. The analyses on the complete cases yielded quite similar results. Prevalence of LBP or LBP patterns differed less than 2% between complete and imputed cases. 4. Discussion 4.1. Main findings Our data show that population prevalence of LBP in the general population is relatively stable over a 10 year period. Despite stable prevalences, LBP is characterized as a very dynamic condition. Six percent of the population suffered from persistent LBP over the
Table 3 Sociodemographic and lifestyle characteristics of participants in three patterns of LBP including the results of the chi-square test. Never longstanding LBP (%) Sociodemographic characteristics Gender Men 50.1 Age 26–35 year 18.6 36–45 year 33.7 46–55 year 28.2 56–65 year 19.5 Educational level
Work status
Varying LBPa (%)
Overall test
41.4 9.0 30.5 32.0 28.5
42.2 17.0 31.7 31.7 19.6
0.00 0.00
Low
51.3
66.5
59.3
0.00
Medium High Paid job
27.7 21.0 64.8
19.6 13.9 41.2
25.3 15.4 57.9
0.00
50.9 39.5 9.6
41.4 40.5 18.1
47.6 41.2 11.2
0.00
29.0 77.8
41.2 73.7
33.6 75.1
0.00 0.30
Lifestyle characteristics BMI <25 Overweight Obesity Smoking Physical activity
Persistent LBP over 10 years (%)
Smoker Sufficient
a Varying LBP consisted of new episodes of long-standing LBP, recovery of longstanding LBP, and a variable course of long-standing LBP.
complete follow-up. Sixty-two did not mention long-standing LBP over a 10-year period and the rest demonstrated a varying course of LBP, with equal proportions of participants who recovered, reported a new episode or a variable pattern of long-standing LBP. We found that individuals with persistent LBP differ in many respects from participants who never experienced long-standing LBP. Participants with persistent LBP and to a lesser extent those with a varying pattern of LBP, are older, lower educated, female, smokers, overweight and report more frequently not having a paid job than those without long-standing LBP. Ten-year persistent LBP, both in the general population and among those with long-standing LBP at baseline, is consistently associated with higher age, not having a paid job and smoking in multivariate models. Women and smokers are at increased risk for a new episode of long-standing LBP, but the associations that we found are considered to be weak. 4.2. Comparison of literature Few other prospective cohort studies investigated patterns of LBP over time in a general population (Elliott et al., 2002; Waxman et al., 2000). A population-based study with a follow-up of 3 years and two measurements found that 29% had persistent LBP while 31% did not report LBP during lifetime (Waxman et al., 2000). The remaining 40% had varying patterns of LBP. Elliott et al. found that 36% had persistent LBP while another 36% did not have longstanding LBP during a 4 year follow-up (Elliott et al., 2002). Kääriä et al. investigated LBP in a cohort of industrial employees over a period of 28 years (Kaaria et al., 2006). The prevalence of persistent LBP was 31% in this study, whereas only 11% never reported LBP during this period. The frequency of persistent LBP in those studies is higher than we found. Variation in defining long-standing LBP, or even the use of acute LBP (Waxman et al., 2000), might explain this difference. For example, Kääriä et al. included ache, stiffness, sensitivity to movement, and numbness of the low back instead of just LBP (Kaaria et al., 2006). Furthermore, in two of those studies only two measures of long-standing LBP were performed (Elliott et al., 2002; Waxman et al., 2000) which might result in a higher percentage
S.H. van Oostrom et al. / European Journal of Pain 15 (2011) 993–998
997
Table 4 Predictors (t1) of persistent LBP or new episodes of long-standing LBP in a period of 10 years; results of the multivariate logistic regression analyses in the imputed dataset. Model 1 Persistent LBP among total (n = 4738)c
Model 2 Persistent LBP among those with long-standing LBP at t1 (n = 904–912)d
Model 3 New episodes of long-standing LBP among those without long-standing LBP at t1 (n = 3826–3834)d
Sociodemographic characteristics Female vs male 0.91 (0.69 1.20)
0.90 (0.65 1.24)
1.41 (1.18 1.70)
Age (per 10 years) 26–35 year 36–45 year 46–55 year 56–65 year
1.00 1.83 (1.05 3.19) 2.14 (1.30 3.54) 2.17 (1.19 3.98)
1.00 1.93 (1.01 3.69) 1.77 (0.99 3.17) 2.09 (1.03 4.23)
1.00 0.97 (0.75 1.26) 1.02 (0.77 1.35) 0.80 (0.60 1.08)
Educational levela Low Medium High
1.00 0.82 (0.59 1.13) 0.75 (0.51 1.11)
1.00 1.06 (0.71 1.57) 1.21 (0.75 1.95)
1.00 0.94 (0.77 1.15) 0.81 (0.64 1.03)
Work status
1.51 (1.15 1.96)
0.62 (0.41 0.95)
1.00 (0.82 1.23)
Lifestyle characteristics BMI <25 Overweight Obesity
1.00 1.03 (0.78 1.35) 1.62 (1.12 2.34)
1.00 0.92 (0.67 1.28) 1.34 (0.85 2.13)
1.00 1.07 (0.84 1.36) 1.16 (0.89 1.52)
Smoking Physically active vs inactiveb
1.51 (1.15 1.96) 0.87 (0.63 1.20)
1.38 (1.00 1.92) 1.04 (0.66 1.66)
1.27 (1.04 1.56) 0.86 (0.68 1.08)
a
Low = intermediate secondary education or less; Medium = intermediate vocational or higher secondary education; High = higher vocational education or university. P3.5 h per week spent on cycling, gardening or sports activities with MET P 4.0 i.e. Dutch recommendations of healthy physical activity (only available for 4/5 of respondents). c In total, 4738 participants were included in model 1 because measurements of physical activity performed in 1993 (of t1) were excluded. d The analyses on subgroups of our datasets, differ in the number of participants over the five imputation datasets. b
of persistent LBP compared with the use of three measures in our study. Few studies investigated risk factors for long-standing LBP or even for the transition from acute to long-standing LBP (Chou and Shekelle, 2010; Henschke et al., 2008), however, risk factors for persistent LBP over 10 years were not studied earlier. A recent literature overview identified age, smoking, physical activity and obesity as risk factors for long-standing LBP, while physical activity was not associated with long-standing LBP in our study (Krismer and van Tulder, 2007). A recent inception cohort study found that older age was prognostic for long-standing LBP among patients with acute LBP (Henschke et al., 2008). Being older and obesity were not identified as risk factors in a systematic review studying risk factors for the transition from acute to long-standing LBP, whereas smoking was borderline significant (Chou and Shekelle, 2010). Few studies examined associations of demographic and lifestyle characteristics with the development of long-standing LBP (Elliott et al., 2002; Thomas et al., 1999). The finding that women are at slightly higher risk for a new episode of long-standing LBP is not supported by a 4-year follow-up study on the course of LBP (Elliott et al., 2002). Whereas a 1-year follow-up study supported the higher risk for smokers and women and also found a higher risk for being older (Thomas et al., 1999). 4.3. Methodological considerations The main strength of this study is the longitudinal prospective design of the Doetinchem Cohort Study. Information about LBP was available from three measurements which covered a period of 10 years. We were able to study LBP patterns over this period, which is unique in this research area. The Doetinchem Cohort Study is a population-based study conducted in a rural area of the Netherlands. In general, cohort data like the Doetinchem Cohort Study show a slightly too rosy picture on health and health determinants, such as the physical activity levels in the study. Physical activity levels are high compared with average levels of
physical activity in the Netherlands due to high levels of cycling, gardening and heavy work. Although the participation rate in this prospective study is relatively high, dropout is inevitable. The Doetinchem Cohort Study comprised selective non-response because non-respondents are more frequently overweight, lower educated, smokers, and inactive compared to complete cases (Blokstra et al., 2010). Therefore, multiple imputations was used to tackle the problem of nonresponse. A limitation of our study is that long-standing LBP was selfreported and not measured exact the same over the three measurements. Long-standing LBP was defined as pain for more than 1 month at t2, which is not a common definition of long-standing or chronic LBP (de Vet et al., 2002). Since long-standing LBP is usually defined as LBP longer than 3 months, the number of participants with long-standing LBP may be overestimated at t2 but the data shows that the percentage participants with long-standing LBP at t2 was not higher than at t1 and t3. This showed that the differences in definitions result in only small differences in the group of LBP sufferers. In addition, the measurement of LBP in the general population is different from the LBP presented at a GP or sick leave due to LBP. In general, those found with LBP in the general population at one moment in time are those with long-standing pain and therefore the difference between 1 and 3 months is negligible. Furthermore, the reverse of the unique long-term follow-up in this study is that LBP was measured once in 5 years and information concerning the remaining 4 years between two measurements is missing. Owing to the recurrent nature of LBP it is likely that the LBP status changes in 4 years. Therefore, future studies investigating patterns of LBP should include both valid and consistent measures and also more frequent measures to improve the reliability of patterns. 4.4. Implications About 6% of the general population has persistent LBP over 10 years, this can be regarded as the most severe subgroup of
998
S.H. van Oostrom et al. / European Journal of Pain 15 (2011) 993–998
LBP patients. A much larger group of LBP patients shows a varying pattern of LBP. The dynamic character of LBP is a challenge for epidemiological studies. Many studies on the recovery of LBP have follow-up periods of less than 1 or 2 years. Actually, our long-term data imply that recovery is often a temporary state in a varying pattern of LBP. Future studies are recommended to carefully consider LBP outcomes with regard to the length of the follow-up. The recovery of LBP should therefore not be the only outcome of intervention studies; also attention could be paid to prevention or postponement of recurrent LBP. Furthermore, individuals with persistent LBP and varying LBP scored more unfavorable on lifestyle risk factors than individuals without LBP. This implies that an unhealthy lifestyle is not only a risk factor for well known diseased as cancer, cardiovascular diseases and diabetes, but affect the occurrence of persistent LBP as well. Those who are older, obese, do smoke and do not have a paid job can be regarded as high risk groups for persistent LBP and might deserve extra attention for interventions, in particular to improvement or maintenance of functioning and participation, since it is not likely that the LBP of these individuals will disappear. Acknowledgements The Doetinchem Cohort Study was financially supported by the Ministry of Health, Welfare and Sport of the Netherlands and the National Institute for Public Health and the Environment. We thank the epidemiologists and fieldworkers of the Municipal Health Service in Doetinchem for their contribution to the data collection for this study. References Allison PD. Missing data. Series Sup, editor. Thousand Oaks, CA: Sage; 2001. Bjorck-van Dijken C, Fjellman-Wiklund A, Hildingsson C. Low back pain, lifestyle factors and physical activity: a population based-study. J Rehabil Med 2008;40:864–9. Blokstra A, Smit HA, Verschuren WMM. Veranderingen in leefstijl – en risicofactoren voor chronische ziekten met het ouder worden: De Doetinchem Studie 1987–2002 (in Dutch). Bilthoven: National Institute of Public Health and the Environment; 2010. Chou R, Shekelle P. Will this patient develop persistent disabling low back pain? JAMA 2010;303:1295–302. de Vet HC, Heymans MW, Dunn KM, Pope DP, van der Beek AJ, Macfarlane GJ, et al. Episodes of low back pain: a proposal for uniform definitions to be used in research. Spine 2002;27:2409–16. Dunn KM, Jordan K, Croft PR. Characterizing the course of low back pain: a latent class analysis. Am J Epidemiol 2006;163:754–61. Elliott AM, Smith BH, Hannaford PC, Smith WC, Chambers WA. The course of chronic pain in the community: results of a 4-year follow-up study. Pain 2002;99:299–307. Freburger JK, Holmes GM, Agans RP, Jackman AM, Darter JD, Wallace AS, et al. The rising prevalence of chronic low back pain. Arch Intern Med 2009;169:251–8. Hagen KB, Thune O. Work incapacity from low back pain in the general population. Spine 1998;23:2091–5.
Harkness EF, Macfarlane GJ, Silman AJ, McBeth J. Is musculoskeletal pain more common now than 40 years ago?: two population-based cross-sectional studies. Rheumatology 2005;44:890–5. Hayden JA, Dunn KM, van der Windt DA, Shaw WS. What is the prognosis of back pain? Best Pract Res Clin Rheumatol 2010;24:167–79. Henschke N, Maher CG, Refshauge KM, Herbert RD, Cumming RG, Bleasel J, et al. Prognosis in patients with recent onset low back pain in Australian primary care: inception cohort study. Bmj 2008;337:171. Hestbaek L, Leboeuf-Yde C, Kyvik KO. Are lifestyle-factors in adolescence predictors for adult low back pain? A cross-sectional and prospective study of young twins. BMC Musculoskelet Disord 2006a;7:27. Hestbaek L, Leboeuf-Yde C, Kyvik KO, Manniche C. The course of low back pain from adolescence to adulthood: eight-year follow-up of 9600 twins. Spine 2006b;31:468–72. Hestbaek L, Leboeuf-Yde C, Manniche C. Low back pain: what is the long-term course? A review of studies of general patient populations. Eur Spine J 2003;12:149–65. Huppe A, Muller K, Raspe H. Is the occurrence of back pain in Germany decreasing? Two regional postal surveys a decade apart. Eur J Public Health 2007;17:318–22. Kaaria S, Luukkonen R, Riihimaki H, Kirjonen J, Leino-Arjas P. Persistence of low back pain reporting among a cohort of employees in a metal corporation: a study with 5-, 10-, and 28-year follow-ups. Pain 2006;120:131–7. Klebanoff MA, Cole SR. Use of multiple imputation in the epidemiologic literature. Am J Epidemiol 2008;168:355–7. Krismer M, van Tulder M. Strategies for prevention and management of musculoskeletal conditions. Low back pain (non-specific). Best Pract Res Clin Rheumatol 2007;21:77–91. Kuorinka I, Jonsson B, Kilbom A, Vinterberg H, Biering-Sorensen F, Andersson G, et al. Standardised Nordic questionnaires for the analysis of musculoskeletal symptoms. Appl Ergon 1987;18:233–7. Leijon O, Mulder M. Prevalence of low back pain and concurrent psychological distress over a 16-year period. Occup Environ Med 2009;66:137–9. Leino PI, Berg MA, Puska P. Is back pain increasing? Results from national surveys in Finland during 1978/9–1992. Scand J Rheumatol 1994;23:269–76. Mitchell T, O’Sullivan PB, Burnett A, Straker L, Smith A, Thornton J, et al. Identification of modifiable personal factors that predict new-onset low back pain: a prospective study of female nursing students. Clin J Pain 2010;26:275–83. Picavet HS, Struijs JN, Westert GP. Utilization of health resources due to low back pain: survey and registered data compared. Spine 2008;33:436–44. Puts MT, Deeg DJ, Hoeymans N, Nusselder WJ, Schellevis FG. Changes in the prevalence of chronic disease and the association with disability in the older Dutch population between 1987 and 2001. Age Ageing 2008;37:187–93. Rubin DI. Epidemiology and risk factors for spine pain. Neurol Clin 2007;25:353–71. Schafer JL. Analysis of incomplete multivariate data. London: Chapman and Hall; 1997. Sterne JA, White IR, Carlin JB, Spratt M, Royston P, Kenward MG, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. Bmj 2009;338:b2393. Thomas E, Silman AJ, Croft PR, Papageorgiou AC, Jayson MI, Macfarlane GJ. Predicting who develops chronic low back pain in primary care: a prospective study. Bmj 1999;318:1662–7. van Buuren S, Groothuis-Oudshoorn K. MICE; Multivariate Imputation by Chained Equations in R. J Stat Softw, in press. van Tulder MW, Koes BW, Metsemakers JF, Bouter LM. Chronic low back pain in primary care: a prospective study on the management and course. Fam Pract 1998;15:126–32. Verschuren WM, Blokstra A, Picavet HS, Smit HA. Cohort profile: the Doetinchem Cohort Study. Int J Epidemiol 2008;37:1236–41. Von Korff M, Saunders K. The course of back pain in primary care. Spine 1996;21:2833–7 [discussion 8–9]. Waxman R, Tennant A, Helliwell P. A prospective follow-up study of low back pain in the community. Spine 2000;25:2085–90.
Contents lists available at ScienceDirect
European Journal of Pain journal homepage: www.EuropeanJournalPain.com
Ten year course of low back pain in an adult population-based cohort – The Doetinchem Cohort Study Sandra H. van Oostrom a,⇑, W.M. Monique Verschuren a, Henrica C.W. de Vet b, H. Susan J. Picavet a a b
National Institute of Public Health and the Environment, Bilthoven, The Netherlands Department of Epidemiology and Biostatistics and the EMGO Institute for Health and Care Research, VU University Medical Centre, Amsterdam, The Netherlands
a r t i c l e
i n f o
Article history: Received 2 August 2010 Received in revised form 18 January 2011 Accepted 24 February 2011 Available online 22 March 2011 Keywords: Low back pain Persistence Prospective cohort Prevalence Course
a b s t r a c t Background: The long-term course of long-standing low back pain is largely unknown since long-term data are scarce. Objective: We examined the course of self-reported low back pain in the prospective population-based Doetinchem cohort over a period of 10 years. Methods: Between 1993 and 2007, around 5700 randomly selected men and women in four age groups of originally 20–29, 30–39, 40–49, 50–59 years were measured three times. Logistic regression analysis was used to study the association of sociodemographic (gender, age, education, work status) and lifestyle characteristics (BMI, smoking, physical activity) with persistent and new episodes of long-standing low back pain. Results: The prevalence of long-standing low back pain is quite stable over a 10 year period, approximately 20% on population level. On individual level, around 30% of the population was completely free of low back pain during the entire period, 6% can be characterized as persistent back pain sufferers. Individuals with persistent and a varying pattern have a more unhealthy lifestyle (BMI and smoking) than those without low back pain. Age, smoking, obesity and not having a paid job are associated with 10-year persistent back pain in the general population, whereas age and not having a paid job are associated in those with long-lasting back pain at baseline. New episodes of long-standing back pain are relatively frequent among women and smokers. Conclusions: Low back pain in the population is characterized as very dynamic which challenges epidemiological studies highly. Long-term information on the course of back pain is needed to define severe subgroups. Ó 2011 European Federation of International Association for the Study of Pain Chapters. Published by Elsevier Ltd. All rights reserved.
1. Introduction Low back pain (LBP) is a major health problem in developed countries; more than 70% of the population will experience an episode of LBP at some time in their lives (Rubin, 2007). Estimates of prevalence of LBP over longer periods of time showed an increase (Freburger et al., 2009; Harkness et al., 2005; Leijon and Mulder, 2009), a decrease (Puts et al., 2008; Huppe et al., 2007) or a stable prevalence over time (Leino et al., 1994). The course of LBP is characterized by variation and change (Von Korff and Saunders, 1996). Acute LBP generally resolves within weeks but recurrences and long-term persistence of the pain are ⇑ Corresponding author. Address: Department for Prevention and Health Services Research (PZO, pb 101), National Institute of Public Health and the Environment, PO box 1, 3720 BA Bilthoven, The Netherlands. Tel.: +31 30 2748596; fax: +31 30 2744407. E-mail address: [email protected] (S.H. van Oostrom).
common (Hestbaek et al., 2006b; Dunn et al., 2006). Recent estimates showed an increase of chronic LBP to 10% of the population in the US (Freburger et al., 2009). Individuals with chronic LBP report high pain levels, high health care utilization and frequent work disability (van Tulder et al., 1998; Hagen and Thune, 1998; Picavet et al., 2008). Chronic LBP is therefore referred to as a severe and costly condition. However, the long-term course of chronic LBP is largely unknown. It is suggested that individuals with chronic, long-standing LBP, which is usually defined as LBP for more than 3 months, tend to show a more persistent course (Hayden et al., 2010). However, even in long-standing LBP pain might not persist during the entire life course. Follow-up periods of most prospective studies investigating the course of LBP are limited to 1 year, which can be regarded as relatively short for a dynamic condition as LBP (Dunn et al., 2006). Understanding the long-term course of LBP is important because it provides information on the need for prevention and treatment. Life long prospective studies in the general population are
1090-3801/$36.00 Ó 2011 European Federation of International Association for the Study of Pain Chapters. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ejpain.2011.02.007
994
S.H. van Oostrom et al. / European Journal of Pain 15 (2011) 993–998
regarded the optimal method for studying the course of LBP (Hestbaek et al., 2003). Unfortunately, life long prospective studies are scarce. The ability to predict which individuals are susceptible to long-standing LBP would be a major step ahead with regard to prevention. Recent studies point out that unhealthy lifestyles such as smoking and physical inactivity are associated with LBP (Hestbaek et al., 2006a; Mitchell et al., 2010; Bjorck-van Dijken et al., 2008). Lifestyle-factors might be important elements of prevention, since they are modifiable in nature. Yet, the role of lifestyle factors for long-standing LBP is unclear. The objective of the current study is to investigate the course of self-reported LBP over 10 years in the Doetinchem Cohort Study, a prospective study in the general population in the Netherlands. Sociodemographic and lifestyle characteristics of LBP patterns over time were investigated. Furthermore, we studied whether 10-year persistent LBP and new episodes of long-standing LBP can be predicted based on simple sociodemographic and lifestyle characteristics. 2. Methods 2.1. Study population The Doetinchem Cohort Study is a prospective observational longitudinal population-based study with 15 years of follow-up. First measurements (t0) took place between 1987 and 1991 (see flow chart in Fig. 1). In that period 12,405 inhabitants of Doetinchem between 20 and 60 years old were examined as part of the ‘Monitoring Project on Cardiovascular Disease Risk Factors’ (MP-CVDRF). From the participants of the first examination (t0), a random sample of 7769 was invited to participate for a second examination in the Monitoring Project on Chronic Disease Risk Factors (MORGEN-project) (t1: 1993–1997), of which 79% (n = 6118) participated (Verschuren et al., 2008). From the second examination onwards a measurement of LBP was included. Therefore, the second examination forms the baseline (t1) of our analyses. Those
Round 1, t0 1987 1991 1987-1991
In Doetinch Doetinchem 20,155 20 155 adults 20-59 yrs were invited 12,405 12 405 (62%) invited, were examined
Round 2, t1 1993-1997, baseline
7,769 were invited, 6,118 i d (79%) were examined
Round 3, t2 1998-2002 1
6,579 were invited, 4,917 (75%) were examined
Round 4, t3 2003-2007
5,783 were invited, 4,520 (78%) were examined
invited for the second examination, were invited for the follow-up measurements 5 and 10 years later for a third (t2: 1998–2002) and fourth examination (t3: 2003–2007). Participants who actively refused to participate in the second or third examination were not invited again. All participants who were invited for t3 and participated in t1, t2 or t3 were selected for this study, ending up with a total of 5706 participants. The study was approved by the Medical Ethics Committee of the Netherlands Organization of Applied Scientific Research Institute. All participants gave written informed consent. 2.2. Data collection and processing Measurements included questionnaires and a physical examination. Details on sampling and data collection procedures are described elsewhere (Verschuren et al., 2008). Information about sociodemographic characteristics, lifestyle factors and medical history of (chronic) diseases was obtained by self-report. Sociodemographic and lifestyle factors as reported at baseline were used for this study. Age was categorized in four 10-year categories: 26–35, 36–45, 46–55, and 56–65 years old. Educational level was assessed as the highest level reached and classified into three categories: low (intermediate secondary education or less), medium (intermediate vocational or higher secondary education) and high (higher vocational education or university). Smoking was categorized as current smoker or non-smoker, the latter included ex-smokers. Physical activity was assessed by an extensive physical activity questionnaire, from which the total time spent on moderate-to-vigorous physical activities was calculated. Those who met the recommended level of 3.5 h a week spent on at least moderately intense physical activity were categorized as being active, and those who did not as being inactive. Body mass index (BMI) was calculated from measured height (m) and body weight (kg). Overweight was defined as a BMI between 25 and 30 kg/m2, and obesity was defined as a BMI equal to or above 30 kg/m2.
This study consisted of 5,706 participants who were invited for t3 and participated i t1 in t1, t2 or t3 t3. The complete case anal analysis sis consisted of 4,007 participants who completed all three examinations. examinations
Fig. 1. Flow chart of the Doetinchem Cohort Study.
S.H. van Oostrom et al. / European Journal of Pain 15 (2011) 993–998
2.3. Measurement of low back pain The 12-month period prevalence of LBP and long-standing LBP were measured at t1, t2 and t3. The 12-month prevalence of LBP was measured with the question ‘Have you had trouble, discomfort or pain in the lower part of the back during the last 12 months?’ that was based on the Nordic back questionnaire (Kuorinka et al., 1987). This question was followed by questions regarding LBP at the moment, the number of episodes of LBP in the past year and the duration of LBP. These questions were used to create a measure for long-standing LBP, which is the main focus of this study. The question about the duration of LBP was formulated like ‘In the past 12 months, how long in total did you have LBP?’, with the following seven response options ‘less than a week’, ‘1 or 2 weeks’, ‘3 or 4 weeks’, ‘5 or 6 weeks’, ‘7–12 weeks’, ‘more than 12 weeks’ and ‘pain is always present’. Long-standing LBP was assigned at t1 when there was one episode of LBP lasting more than 12 weeks or when pain is always present during the past year. At t2 the question about the duration of LBP was formulated like ‘How long have you had LBP for at the moment?’ with the following three response options ‘a few days’, ‘less than a month’ and ‘more than a month’. At t3 the question about the duration of LBP was formulated like ‘Has your current LBP lasted for more than 3 months?’. Due to different phrasing at the measurements different cutoff points were used to define whether LBP was long-standing: at t1 and t3 the cutoff point was more than 3 months, at t2 more than 1 month. The 12-month period prevalence of LBP and the prevalence of long-standing LBP at the three rounds were calculated. Five patterns of long-standing LBP were composed based on the three measurements: (1) no long-standing LBP = no long-standing LBP at t1, t2 and t3; (2) new episode of long-standing LBP = no long-standing LBP at t1 and long-standing LBP at t2 or t3; (3) recovery of long-standing LBP = long-standing LBP at t1 and no long-standing LBP at t2 or t3; (4) variable long-standing LBP = the reporting of long-standing LBP varied each subsequent measurement; (5) persistent LBP = long-standing LBP at t1, t2 and t3.
995
the measurement of physical activity changed after this year. All statistical analyses were conducted in SAS software version 9.1.
2.5. Missing data and multiple imputations Age and sex were available for all participants in our study. About 4.6% of the participants did not participate at t1, 15.8% and 20.8% of the participants had complete missing data at t2 and t3. Complete case analysis, including participants who participated in all three measurements, may lead to biased results since data are unlikely to be ‘missing completely at random’(Sterne et al., 2009; Klebanoff and Cole, 2008). Data are ‘missing completely at random’ when no systematic differences exist between the missing values and observed values. When systematic differences exist that can be explained by differences in observed data the data are ‘missing at random’. At t2 and t3, responders have been compared to the non-responders with respect to several variables (Blokstra et al., 2010). Compared to responders, non-responders were lower educated, obese, smokers, and less active. Therefore, we assumed that our data are ‘missing at random’ and used the Multivariate Imputation by Chained Equations (MICE) package in R to perform multiple imputation (van Buuren and Groothuis-Oudshoorn, in press). MICE allows dichotomous, continuous, and categorized variables. In the MICE procedure logistic regression is applied to impute incomplete dichotomous variables, linear regression to impute continuous variables, and predictive mean matching to impute categorized variables. The imputation matrix consisted of all sociodemographic and lifestyle characteristics at baseline, and the LBP variables at baseline, t2 and t3. Five imputation sets were created, analyzed by logistic regression (as described above), and pooled by the MIANALYZE procedure in SAS. The chi-square tests were pooled by the combchi macro (Schafer, 1997; Allison, 2001). Baseline measurements of complete and imputed cases were compared based on means and frequencies. Results from the complete cases and imputed logistic regression analyses were compared and imputed analyses are presented.
3. Results 2.4. Statistical analyses First, the frequency and percentage for each of the LBP patterns were calculated. Second, differences in sociodemographic and lifestyle characteristics were presented and tested by chi-square statistics for (1) no long-standing LBP, (2) persistent LBP over 10 years and (3) varying (episodic) LBP which consisted of a new episode of long-standing LBP, recovery of long-standing LBP and variable long-standing LBP over the measurements. Third, to identify high risk groups for long-standing LBP we determined which sociodemographic and lifestyle characteristics were associated with 10-year persistent and new episodes of long-standing LBP in three logistic regression models: (1) persistent LBP over 10 years in the general adult population; (2) persistent LBP over 10 years among those with long-standing LBP at baseline (t1); (3) new episodes of long-standing LBP in 10 years among those without long-standing LBP at baseline (t1). Sociodemographic characteristics (gender, age, educational level, and work status) and lifestyle factors (body mass category, smoking status, and physical activity) at t1 were included in the logistic regression analyses as potential determinants of persistent or new episodes of LBP. For the logistic regression analyses, 968 respondents who were measured in 1993 of t1 were excluded because
The study population consists of 2686 men and 3020 women aged 26–65 years (mean (SD) 45.9 (10.0)) at baseline. Characteristics of the complete cases (n = 4007) and all participants including imputed cases (n = 5706) are shown in Table 1. Twelve-month period prevalence of LBP was reported by 45% at baseline, while long-standing LBP was reported by 20% respectively (Table 1). On population level, the prevalences of LBP and long-standing LBP remained quite stable over the 10 year period. About 71% ever reported any LBP in three measurements over a period of 10 years, indicating that 29% of the population was free of LBP during the three response periods. Among those reporting any LBP, half of them (37%) reported an episode of long-standing LBP. Table 2 shows the patterns of long-standing LBP over 10 years. About 6% can be characterized as persistent LBP sufferers, 62% had no period of long-standing LBP during the 10 year follow-up, 10% recovered from long-standing LBP, 11% reported a new episode of long-standing LBP, and another 11% had a variable pattern of long-standing LBP. Differences between the patterns no longstanding LBP, persistent LBP and varying LBP were found on all sociodemographic and lifestyle characteristics, except for physical activity. Compared to those without LBP, participants with persistent LBP and a varying pattern of LBP were more often female, older, lower educated, smoker, more frequently overweight and obese, and not having a paid job (Table 3).
996
S.H. van Oostrom et al. / European Journal of Pain 15 (2011) 993–998
Table 1 Sociodemographic characteristics, lifestyle characteristics and LBP at t1 for complete cases and imputed cases. Mean and standard deviation are presented for continuous variables, otherwise percentages are presented. Complete cases (n = 4007) (%)
Imputed cases (n = 5706) (%)
Sociodemographic characteristics Age Men
45.8 (9.8) 47.4
45.9 (10.0) 47.1
Educational level Low Medium High Paid job
51.0 27.4 21.6 63.9
54.7 26.5 18.8 61.3
Lifestyle characteristics Smoking Physical active BMI
27.6 78.9 25.3 (3.6)
31.2 76.7 25.5 (3.7)
Low back pain 12-Month period prevalence (t1) 12-Month period prevalence (t2) 12-Month period prevalence (t3) Long-standing pain (t1) Long-standing pain (t2) Long-standing pain (t3)
45.3 49.0 46.6 17.4 17.4 19.9
45.7 49.4 47.1 18.7 18.4 20.6
LBP patterns over 10 years Ever LBP Ever long-standing LBP Persistent LBP over 10 years New episode of long-standing LBP
70.4 35.7 4.6 18.3
70.6 37.1 5.3 18.3
Table 2 Long-standing LBP patterns of the participants of the Doetinchem Cohort Study over a period of 10 years after multiple imputation. Long-standing LBP
Percentage (n = 5706) (%)
t1
t2
t3
No long-standing LBP New episode of long-standing LBP
No No No
No Yes No
No Yes Yes
62.4 10.8
Recovery of long-standing LBP
Yes Yes
Yes No
No No
10.3
Variable course of long-standing LBP
Yes No
No Yes
Yes No
10.9
Persistent LBP over 10 years
Yes
Yes
Yes
5.6%
Ten-year persistent LBP is predicted by older age, smoking, obesity and not having a paid job at baseline (Table 4, model 1). Among those with long-standing LBP at baseline those with persistent LBP were predicted by being older and smoker, and not having a paid job (model 2). Model 3 investigated the predictors for new episodes of long-standing LBP among participants without longstanding LBP at t1. Being a woman and smoker were predictive for a new episode of long-standing LBP at the 10-year follow-up. The analyses on the complete cases yielded quite similar results. Prevalence of LBP or LBP patterns differed less than 2% between complete and imputed cases. 4. Discussion 4.1. Main findings Our data show that population prevalence of LBP in the general population is relatively stable over a 10 year period. Despite stable prevalences, LBP is characterized as a very dynamic condition. Six percent of the population suffered from persistent LBP over the
Table 3 Sociodemographic and lifestyle characteristics of participants in three patterns of LBP including the results of the chi-square test. Never longstanding LBP (%) Sociodemographic characteristics Gender Men 50.1 Age 26–35 year 18.6 36–45 year 33.7 46–55 year 28.2 56–65 year 19.5 Educational level
Work status
Varying LBPa (%)
Overall test
41.4 9.0 30.5 32.0 28.5
42.2 17.0 31.7 31.7 19.6
0.00 0.00
Low
51.3
66.5
59.3
0.00
Medium High Paid job
27.7 21.0 64.8
19.6 13.9 41.2
25.3 15.4 57.9
0.00
50.9 39.5 9.6
41.4 40.5 18.1
47.6 41.2 11.2
0.00
29.0 77.8
41.2 73.7
33.6 75.1
0.00 0.30
Lifestyle characteristics BMI <25 Overweight Obesity Smoking Physical activity
Persistent LBP over 10 years (%)
Smoker Sufficient
a Varying LBP consisted of new episodes of long-standing LBP, recovery of longstanding LBP, and a variable course of long-standing LBP.
complete follow-up. Sixty-two did not mention long-standing LBP over a 10-year period and the rest demonstrated a varying course of LBP, with equal proportions of participants who recovered, reported a new episode or a variable pattern of long-standing LBP. We found that individuals with persistent LBP differ in many respects from participants who never experienced long-standing LBP. Participants with persistent LBP and to a lesser extent those with a varying pattern of LBP, are older, lower educated, female, smokers, overweight and report more frequently not having a paid job than those without long-standing LBP. Ten-year persistent LBP, both in the general population and among those with long-standing LBP at baseline, is consistently associated with higher age, not having a paid job and smoking in multivariate models. Women and smokers are at increased risk for a new episode of long-standing LBP, but the associations that we found are considered to be weak. 4.2. Comparison of literature Few other prospective cohort studies investigated patterns of LBP over time in a general population (Elliott et al., 2002; Waxman et al., 2000). A population-based study with a follow-up of 3 years and two measurements found that 29% had persistent LBP while 31% did not report LBP during lifetime (Waxman et al., 2000). The remaining 40% had varying patterns of LBP. Elliott et al. found that 36% had persistent LBP while another 36% did not have longstanding LBP during a 4 year follow-up (Elliott et al., 2002). Kääriä et al. investigated LBP in a cohort of industrial employees over a period of 28 years (Kaaria et al., 2006). The prevalence of persistent LBP was 31% in this study, whereas only 11% never reported LBP during this period. The frequency of persistent LBP in those studies is higher than we found. Variation in defining long-standing LBP, or even the use of acute LBP (Waxman et al., 2000), might explain this difference. For example, Kääriä et al. included ache, stiffness, sensitivity to movement, and numbness of the low back instead of just LBP (Kaaria et al., 2006). Furthermore, in two of those studies only two measures of long-standing LBP were performed (Elliott et al., 2002; Waxman et al., 2000) which might result in a higher percentage
S.H. van Oostrom et al. / European Journal of Pain 15 (2011) 993–998
997
Table 4 Predictors (t1) of persistent LBP or new episodes of long-standing LBP in a period of 10 years; results of the multivariate logistic regression analyses in the imputed dataset. Model 1 Persistent LBP among total (n = 4738)c
Model 2 Persistent LBP among those with long-standing LBP at t1 (n = 904–912)d
Model 3 New episodes of long-standing LBP among those without long-standing LBP at t1 (n = 3826–3834)d
Sociodemographic characteristics Female vs male 0.91 (0.69 1.20)
0.90 (0.65 1.24)
1.41 (1.18 1.70)
Age (per 10 years) 26–35 year 36–45 year 46–55 year 56–65 year
1.00 1.83 (1.05 3.19) 2.14 (1.30 3.54) 2.17 (1.19 3.98)
1.00 1.93 (1.01 3.69) 1.77 (0.99 3.17) 2.09 (1.03 4.23)
1.00 0.97 (0.75 1.26) 1.02 (0.77 1.35) 0.80 (0.60 1.08)
Educational levela Low Medium High
1.00 0.82 (0.59 1.13) 0.75 (0.51 1.11)
1.00 1.06 (0.71 1.57) 1.21 (0.75 1.95)
1.00 0.94 (0.77 1.15) 0.81 (0.64 1.03)
Work status
1.51 (1.15 1.96)
0.62 (0.41 0.95)
1.00 (0.82 1.23)
Lifestyle characteristics BMI <25 Overweight Obesity
1.00 1.03 (0.78 1.35) 1.62 (1.12 2.34)
1.00 0.92 (0.67 1.28) 1.34 (0.85 2.13)
1.00 1.07 (0.84 1.36) 1.16 (0.89 1.52)
Smoking Physically active vs inactiveb
1.51 (1.15 1.96) 0.87 (0.63 1.20)
1.38 (1.00 1.92) 1.04 (0.66 1.66)
1.27 (1.04 1.56) 0.86 (0.68 1.08)
a
Low = intermediate secondary education or less; Medium = intermediate vocational or higher secondary education; High = higher vocational education or university. P3.5 h per week spent on cycling, gardening or sports activities with MET P 4.0 i.e. Dutch recommendations of healthy physical activity (only available for 4/5 of respondents). c In total, 4738 participants were included in model 1 because measurements of physical activity performed in 1993 (of t1) were excluded. d The analyses on subgroups of our datasets, differ in the number of participants over the five imputation datasets. b
of persistent LBP compared with the use of three measures in our study. Few studies investigated risk factors for long-standing LBP or even for the transition from acute to long-standing LBP (Chou and Shekelle, 2010; Henschke et al., 2008), however, risk factors for persistent LBP over 10 years were not studied earlier. A recent literature overview identified age, smoking, physical activity and obesity as risk factors for long-standing LBP, while physical activity was not associated with long-standing LBP in our study (Krismer and van Tulder, 2007). A recent inception cohort study found that older age was prognostic for long-standing LBP among patients with acute LBP (Henschke et al., 2008). Being older and obesity were not identified as risk factors in a systematic review studying risk factors for the transition from acute to long-standing LBP, whereas smoking was borderline significant (Chou and Shekelle, 2010). Few studies examined associations of demographic and lifestyle characteristics with the development of long-standing LBP (Elliott et al., 2002; Thomas et al., 1999). The finding that women are at slightly higher risk for a new episode of long-standing LBP is not supported by a 4-year follow-up study on the course of LBP (Elliott et al., 2002). Whereas a 1-year follow-up study supported the higher risk for smokers and women and also found a higher risk for being older (Thomas et al., 1999). 4.3. Methodological considerations The main strength of this study is the longitudinal prospective design of the Doetinchem Cohort Study. Information about LBP was available from three measurements which covered a period of 10 years. We were able to study LBP patterns over this period, which is unique in this research area. The Doetinchem Cohort Study is a population-based study conducted in a rural area of the Netherlands. In general, cohort data like the Doetinchem Cohort Study show a slightly too rosy picture on health and health determinants, such as the physical activity levels in the study. Physical activity levels are high compared with average levels of
physical activity in the Netherlands due to high levels of cycling, gardening and heavy work. Although the participation rate in this prospective study is relatively high, dropout is inevitable. The Doetinchem Cohort Study comprised selective non-response because non-respondents are more frequently overweight, lower educated, smokers, and inactive compared to complete cases (Blokstra et al., 2010). Therefore, multiple imputations was used to tackle the problem of nonresponse. A limitation of our study is that long-standing LBP was selfreported and not measured exact the same over the three measurements. Long-standing LBP was defined as pain for more than 1 month at t2, which is not a common definition of long-standing or chronic LBP (de Vet et al., 2002). Since long-standing LBP is usually defined as LBP longer than 3 months, the number of participants with long-standing LBP may be overestimated at t2 but the data shows that the percentage participants with long-standing LBP at t2 was not higher than at t1 and t3. This showed that the differences in definitions result in only small differences in the group of LBP sufferers. In addition, the measurement of LBP in the general population is different from the LBP presented at a GP or sick leave due to LBP. In general, those found with LBP in the general population at one moment in time are those with long-standing pain and therefore the difference between 1 and 3 months is negligible. Furthermore, the reverse of the unique long-term follow-up in this study is that LBP was measured once in 5 years and information concerning the remaining 4 years between two measurements is missing. Owing to the recurrent nature of LBP it is likely that the LBP status changes in 4 years. Therefore, future studies investigating patterns of LBP should include both valid and consistent measures and also more frequent measures to improve the reliability of patterns. 4.4. Implications About 6% of the general population has persistent LBP over 10 years, this can be regarded as the most severe subgroup of
998
S.H. van Oostrom et al. / European Journal of Pain 15 (2011) 993–998
LBP patients. A much larger group of LBP patients shows a varying pattern of LBP. The dynamic character of LBP is a challenge for epidemiological studies. Many studies on the recovery of LBP have follow-up periods of less than 1 or 2 years. Actually, our long-term data imply that recovery is often a temporary state in a varying pattern of LBP. Future studies are recommended to carefully consider LBP outcomes with regard to the length of the follow-up. The recovery of LBP should therefore not be the only outcome of intervention studies; also attention could be paid to prevention or postponement of recurrent LBP. Furthermore, individuals with persistent LBP and varying LBP scored more unfavorable on lifestyle risk factors than individuals without LBP. This implies that an unhealthy lifestyle is not only a risk factor for well known diseased as cancer, cardiovascular diseases and diabetes, but affect the occurrence of persistent LBP as well. Those who are older, obese, do smoke and do not have a paid job can be regarded as high risk groups for persistent LBP and might deserve extra attention for interventions, in particular to improvement or maintenance of functioning and participation, since it is not likely that the LBP of these individuals will disappear. Acknowledgements The Doetinchem Cohort Study was financially supported by the Ministry of Health, Welfare and Sport of the Netherlands and the National Institute for Public Health and the Environment. We thank the epidemiologists and fieldworkers of the Municipal Health Service in Doetinchem for their contribution to the data collection for this study. References Allison PD. Missing data. Series Sup, editor. Thousand Oaks, CA: Sage; 2001. Bjorck-van Dijken C, Fjellman-Wiklund A, Hildingsson C. Low back pain, lifestyle factors and physical activity: a population based-study. J Rehabil Med 2008;40:864–9. Blokstra A, Smit HA, Verschuren WMM. Veranderingen in leefstijl – en risicofactoren voor chronische ziekten met het ouder worden: De Doetinchem Studie 1987–2002 (in Dutch). Bilthoven: National Institute of Public Health and the Environment; 2010. Chou R, Shekelle P. Will this patient develop persistent disabling low back pain? JAMA 2010;303:1295–302. de Vet HC, Heymans MW, Dunn KM, Pope DP, van der Beek AJ, Macfarlane GJ, et al. Episodes of low back pain: a proposal for uniform definitions to be used in research. Spine 2002;27:2409–16. Dunn KM, Jordan K, Croft PR. Characterizing the course of low back pain: a latent class analysis. Am J Epidemiol 2006;163:754–61. Elliott AM, Smith BH, Hannaford PC, Smith WC, Chambers WA. The course of chronic pain in the community: results of a 4-year follow-up study. Pain 2002;99:299–307. Freburger JK, Holmes GM, Agans RP, Jackman AM, Darter JD, Wallace AS, et al. The rising prevalence of chronic low back pain. Arch Intern Med 2009;169:251–8. Hagen KB, Thune O. Work incapacity from low back pain in the general population. Spine 1998;23:2091–5.
Harkness EF, Macfarlane GJ, Silman AJ, McBeth J. Is musculoskeletal pain more common now than 40 years ago?: two population-based cross-sectional studies. Rheumatology 2005;44:890–5. Hayden JA, Dunn KM, van der Windt DA, Shaw WS. What is the prognosis of back pain? Best Pract Res Clin Rheumatol 2010;24:167–79. Henschke N, Maher CG, Refshauge KM, Herbert RD, Cumming RG, Bleasel J, et al. Prognosis in patients with recent onset low back pain in Australian primary care: inception cohort study. Bmj 2008;337:171. Hestbaek L, Leboeuf-Yde C, Kyvik KO. Are lifestyle-factors in adolescence predictors for adult low back pain? A cross-sectional and prospective study of young twins. BMC Musculoskelet Disord 2006a;7:27. Hestbaek L, Leboeuf-Yde C, Kyvik KO, Manniche C. The course of low back pain from adolescence to adulthood: eight-year follow-up of 9600 twins. Spine 2006b;31:468–72. Hestbaek L, Leboeuf-Yde C, Manniche C. Low back pain: what is the long-term course? A review of studies of general patient populations. Eur Spine J 2003;12:149–65. Huppe A, Muller K, Raspe H. Is the occurrence of back pain in Germany decreasing? Two regional postal surveys a decade apart. Eur J Public Health 2007;17:318–22. Kaaria S, Luukkonen R, Riihimaki H, Kirjonen J, Leino-Arjas P. Persistence of low back pain reporting among a cohort of employees in a metal corporation: a study with 5-, 10-, and 28-year follow-ups. Pain 2006;120:131–7. Klebanoff MA, Cole SR. Use of multiple imputation in the epidemiologic literature. Am J Epidemiol 2008;168:355–7. Krismer M, van Tulder M. Strategies for prevention and management of musculoskeletal conditions. Low back pain (non-specific). Best Pract Res Clin Rheumatol 2007;21:77–91. Kuorinka I, Jonsson B, Kilbom A, Vinterberg H, Biering-Sorensen F, Andersson G, et al. Standardised Nordic questionnaires for the analysis of musculoskeletal symptoms. Appl Ergon 1987;18:233–7. Leijon O, Mulder M. Prevalence of low back pain and concurrent psychological distress over a 16-year period. Occup Environ Med 2009;66:137–9. Leino PI, Berg MA, Puska P. Is back pain increasing? Results from national surveys in Finland during 1978/9–1992. Scand J Rheumatol 1994;23:269–76. Mitchell T, O’Sullivan PB, Burnett A, Straker L, Smith A, Thornton J, et al. Identification of modifiable personal factors that predict new-onset low back pain: a prospective study of female nursing students. Clin J Pain 2010;26:275–83. Picavet HS, Struijs JN, Westert GP. Utilization of health resources due to low back pain: survey and registered data compared. Spine 2008;33:436–44. Puts MT, Deeg DJ, Hoeymans N, Nusselder WJ, Schellevis FG. Changes in the prevalence of chronic disease and the association with disability in the older Dutch population between 1987 and 2001. Age Ageing 2008;37:187–93. Rubin DI. Epidemiology and risk factors for spine pain. Neurol Clin 2007;25:353–71. Schafer JL. Analysis of incomplete multivariate data. London: Chapman and Hall; 1997. Sterne JA, White IR, Carlin JB, Spratt M, Royston P, Kenward MG, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. Bmj 2009;338:b2393. Thomas E, Silman AJ, Croft PR, Papageorgiou AC, Jayson MI, Macfarlane GJ. Predicting who develops chronic low back pain in primary care: a prospective study. Bmj 1999;318:1662–7. van Buuren S, Groothuis-Oudshoorn K. MICE; Multivariate Imputation by Chained Equations in R. J Stat Softw, in press. van Tulder MW, Koes BW, Metsemakers JF, Bouter LM. Chronic low back pain in primary care: a prospective study on the management and course. Fam Pract 1998;15:126–32. Verschuren WM, Blokstra A, Picavet HS, Smit HA. Cohort profile: the Doetinchem Cohort Study. Int J Epidemiol 2008;37:1236–41. Von Korff M, Saunders K. The course of back pain in primary care. Spine 1996;21:2833–7 [discussion 8–9]. Waxman R, Tennant A, Helliwell P. A prospective follow-up study of low back pain in the community. Spine 2000;25:2085–90.