Prognostic factors for pain chronicity in low back pain: a systematic review

Supplemental Digital Content is Available in the Text. Several prognostic factors are related to low back pain chronicity, and these should be taken into account when planning more comprehensive models in its prevention.


Introduction
Low back pain (LBP) is the leading cause of years lived in disability in high-income and middle-income countries. 39 Moreover, a similar increase has also been seen in low-income countries. 68 In 2015, LBP was responsible for approximately 60.1 million years lived in disabilities, an increase of 54% since 1990. 39 For industrialized countries, LBP is a very costly illness 21,138 and indirect costs (work absenteeism, productivity loss) account for more than half of the total costs. 9 In many patients, the specific nociceptive source of LBP cannot be identified and those affected are often classified as having so-called "nonspecific low back pain." 84 Nonspecific LBP represents 90% to 95% of cases, with other causes being specific spinal pathology (,1% of cases) and radicular syndrome (approximately 5%-10% of cases). 7 The global point prevalence of activitylimiting LBP lasting more than 1 day is estimated to be 12%. 69 Although most patients with acute LBP show rapid improvements in pain and disability within 1 month, 106 between 4% and 25% of patients drift to chronicity. 92 The prevalence of chronic low back pain (CLBP) increases linearly from the third decade of life until the age of 60 years, with CLBP being more prevalent in women. 92 The prognosis of nonspecific LBP is greatly influenced by factors not related to the spine. 115 In 1987, a biopsychosocial model for understanding LBP was first introduced by George Waddell. 136 The idea behind the model is based on how psychologic and social influences modulate an individual's perception of symptoms. An overemphasis on pain alone and a dependence on only mechanical, nominal diagnosis can lead to more disability. Therefore, when treating patients with LBP, clinicians should consider all aspects (biomechanical, psychological, and psychosocial) of the illness.
To date, few comprehensive reviews have studied the risks of chronicity in patients with LBP. A review by Valat et al. in 1997 133 concluded that CLBP is more closely related to demographic, psychosocial, and occupational factors than to the medical characteristics of the disorder itself. A 2010 systematic review of "yellow flag" risk factors for developing CLBP 15 concluded that maladaptive pain coping behaviors, lower functional impairment at baseline, nonorganic signs referring to somatization, worse general health status before the onset of pain, and the presence of psychiatric comorbidities were significant in terms of chronicity.
Since then, a large number of studies have focused on revealing the risk factors behind this global problem.
The aim of this systematic review is to identify the prognostic factors for pain chronicity in patients with LBP and to provide an update on the existing data.

Literature search
Systematic literature searches from computerized databases were conducted until March 30, 2020. The search strategy was developed in collaboration with an information specialist. The following databases were searched without any date restriction: MEDLINE (PubMed), Cochrane Database, and Medic specifically for articles in the Finnish language. The primary target of the search was articles concerning predictive risk factors for chronic, nonspecific LBP. The full search strategy is presented in Appendix 1 (available at http://links.lww.com/PR9/A99).

Study selection and inclusion criteria for selection of studies
The study types included in the literature search were cohort studies, follow-up studies, and reviews. The reviews were used only to search for additional articles to avoid duplication. Randomized controlled trials were not included because the effect of the intervention on the outcome (CLBP) could not be excluded and observing only the group without intervention could create bias. However, studies with interventions could be included if the intervention concerned the whole followed population or its impact could be taken into account in some other way. The references of the studies that met the inclusion criteria were searched for additional articles. There was no time limit for the search. Studies in the English or Finnish languages that focused on working population (aged 18-65 years) were included. If older individuals were recruited, the mean age with SD had to be no more than 65 years. The main outcome was nonspecific CLBP with or without pain radiation, but specific nerve root disorders were excluded. Articles that dealt only with operative treatment were also excluded. Chronic pain is most commonly described as lasting longer than 3 months. 129 Therefore, studies must have assessed the predictive risk factors before that period to be included in the search. A chronic condition was defined as persistent pain in the lower back for a period of 3 months or longer.

Quality assessment
Study quality was assessed using the National Institute of Health study assessment tool. 94 Two independent reviewers evaluated all the included articles according to assessment tool criteria. If the ratings differed, the reviewers discussed the article in an effort to reach consensus. If consensus was not achieved, a third reviewer was consulted. Each study was judged as good, fair, or poor by evaluating the potential risk of bias resulting from the existing flaws.

Results of the search
A Prisma flow chart of the study selection is presented in Figure 1. A total of 2,028 articles were identified. The first exclusion round was based on inappropriate titles or abstracts. We then read the full text of 111 articles, and 25 articles met all the inclusion criteria. Characteristics of the included studies are presented in Table 1. Of these 25 articles, 17 68% were published in 2010 or thereafter. 32,[56][57][58][59][60][61][62][63]83,88,89,97,99,103,119,122 Two articles were found from the references of included articles. 46,55 The excluded articles and the reasons for exclusion are listed in Table 2. Most of the      excluded articles did not meet the criteria concerning the prospective information before the onset of chronic pain, the chronic pain was defined as lasting less than 3 months/12 weeks, or the pain was already chronic at baseline. In some articles concerning the working population, the chronic disease was only defined according to the time spent on sick leave without explaining whether the sick leave was due to LBP or to some other medical condition. In many of the excluded articles, the outcome was defined as timely pain during the follow-up contact compared with persistent symptoms for at least 3 months.

Prognostic risk factors
All prognostic factors are presented in Table 4. In total, 80 prognostic factors were found from the studies.

Personal factors and medical history
Three fair-quality studies found higher body weight to increase the risk of CLBP. 59 and 1 poor-quality study, 110 although statistical significance was achieved only in the latter. There was inconclusive evidence about age as a risk factor, although 2 fair-quality studies 32,55 had a statistically significant result about age being a risk of chronicity. In 2 fair-quality studies, smoking and/or nicotine dependence was statistically significant risk factor. 6,119 The only study rated as good quality found a statistically significant association between higher blood pressure and lower chronicity. 46

Symptom characteristics
Higher pain intensity seemed to increase the risk of CLBP according to 6 studies, 54,55,89,110,122,140 from which statistical significance was achieved in 4. 54,55,122,140 Longer duration of symptoms before the onset of entering the studies (less than 3 months) was found to be predictive for chronicity in 1 fair-quality study. 55 Seven studies investigated functional limitation and disability because of LBP as a risk factor, 19,54,55,88,89,110,140 from which statistical significance was achieved in 1 study. 140

Biomechanical factors
Carrying heavy loads at work was the most studied biomechanical risk factor for chronicity in 3 fair-quality studies 32,56,58 and 2 poor-quality studies, 103,110 and statistically significant in 3. 35,58,83 Other significant factors predicting chronicity with statistical significance according to more than 1 study included particularly physical work 56,58 and difficult working positions. 56,83,103 Furthermore, vibrations and jolts at work significantly increased the risk of chronicity in 1 fair-quality study 56 and nonsignificantly in 1 poor-quality study. 103   79 Including chronic population at baseline Lagersted-Olsen et al. 80 Baseline information inadequate (continued on next page) 6 (2021) e919 www.painreportsonline.com

Psychosocial factors
Numerous psychosocial factors were identified. Depression was the most studied factor predicting chronicity with statistically significant results in 2 studies 55,119 and nonsignificantly in 4. 32,89,110,140 Psychological risk factors that were investigated in more than 1 study included fear avoidance, 54 Compared with previous reviews, 15,133 new factors were found to be predictive of CLBP. Of these, the most evident were obesity, smoking, higher pain intensity, and occupational factors, such as difficult working positions, vibrations, and jolts at work.

Discussion
The main findings in this review are that higher pain intensity, higher body weight, carrying heavy loads at work, difficult working positions, and depression are the most frequently observed Excluded articles with reasons for exclusion.

Article Reason for exclusion
Matsuda et al. 85 Only chronic population at baseline Matsudaira et al. 87 Baseline information inadequate Matsudaira et al. 86 Baseline information inadequate Melloh et al. 90 Different definition for chronic pain; .6 wk, measured by oswestry Mercado et al. 91 Baseline information inadequate, multiple pain sites Neubauer et al. 95 Including chronic population at baseline Nisenzon et al. 98 Baseline information inadequate Noormohammadpour et al. 100 Only chronic population at baseline Nordstoga et al. 101 Only chronic population at baseline Oliveira et al. 102 Only chronic population at baseline Pagé et al. 104 Only chronic population at baseline Picavet et al. 107 Baseline information inadequate Pinheiro et al. 108 Only chronic at baseline Pinto et al. 109 Only chronic population at baseline

Popescu and Lee 111 Dissertation
Rabey et al. 112 Only chronic population at baseline  134 Only chronic population at baseline Werneke et al. 142 Different definition for chronic pain; pain during the past week at follow-up Wilkens et al. 143 Only chronic population at baseline Villafañ e et al. 135 Only chronic population at baseline Williams et al. 144 Different definition for chronic pain; point prevalence at follow-up Yosef et al. 146 Including chronic population at baseline prognostic risk factors for CLBP. Moreover, maladaptive behavior strategies, general anxiety, functional limitation during the episode, smoking, and particularly physical work are also explicitly predictive of chronicity. Most frequently observed protective factors were physical exercise and higher blood pressure.
According to the findings of this review, lifestyle-related factors, such as smoking and obesity, are major risk factors for pain chronicity. Odd ratios for smoking differed between 2.49 (95% confidence interval [CI] 1.15-5.40) 119 and 4.41(95% CI 1.50-12.95). 6 In obesity, odd ratios varied between 1.075 (95% CI 1.023-1.128) 59 and 1.21 (95% CI 1.04-1.41) 97 in women and between 1.091 (95% CI 1.027-1.158) 59 and 1.16 (95% CI 1.05-1.29) 63 in men. In general, the findings about the risk factors of pain chronicity are similar. 120,145 Baseline personal factors concerning poorer general health 18 and functionality 18 were found to be significant risk factors for chronic pain in this review. Conversely, physical well-being 140 and physical exercise 97 were found to protect against chronicity. Poor general health and functionality are coherently interrelated to multimorbidity, which is a major risk factor for general pain chronicity. 24 The same nonmodifiable risk factors, such as age and female sex, found in this review are also found to be risk factors for other chronic pain conditions. 28,41 LBP-induced disability and functional limitation were significant risk factors according to the findings of this review. 140 A study by Wand et al. 140 reported that the correlation coefficient between Roland-Morris Disability Questionnaire and CLBP was 0.48. A similar finding about functional impairment at baseline was reported in a previous review. 15 The lower levels of functionality might be a continuum of a person's lifestyle and behavioral factors. Therefore, avoiding bed rest despite the pain seems even more important.
The physical intensity of work, particularly strenuous physical work, carrying heavy loads, and working in difficult working positions, was related to higher chronicity in this review. 32,56,58,83,103 In a study by Machado and colleagues, 83 the carrying of heavy loads was predictive for CLBP with an odds ratio of 8.0 (95% CI 2. 8-22.6). It is possible therefore that the physical work itself is preventing workers from getting back to work in a timely fashion 125 and thereby contributing to the prolongation of the symptoms.
There is previous strong evidence that cognitive factors, such as attitudes, cognitive style, and fear-avoidance beliefs, are related to the development of pain and disability in patients with back pain. 82 Maladaptive behaviors, such as perceived risk of persistence, 55,88 pain catastrophizing, 88 somatization, 88,89 and coping by ignoring pain, 88 were found to be risk factors in a total of 3 studies. It is not always the case that maladaptive behavior is the first step on the road to chronicity. The prospective designs included in this review would, however, implicate such causality, but one might suggest that fear avoidance, eg, is the immediate result of the pain in the acute phase of LBP, as Linton 82 discussed in his review. Low tolerance of pain was a significant risk factor in this review. 88 The low pain threshold is a complex concept and combines both genetic 124 and psychological aspects. In a study of pain thresholds in patients with chronic pain, there was a correlation between lower pain threshold and depressive tendency and hypochondriac concerns. 75 A previous history of LBP substantially increases the risk of a subsequent new episode. 105 In this review, it was found to be a risk factor in 2 studies. 19,122 Interestingly, we found no evidence of sleep disturbances being a risk factor for chronicity. However, since there is a bidirectional relationship between the intensity of LBP and sleep disturbances, 1 one might assume it would also be a risk factor for CLBP. This would be an interesting hypothesis to study in the future.
So-called "yellow flags" is an umbrella term used to describe psychological risk factors and social and environmental risk factors for prolonged disability and failure to return to work as a consequence of musculoskeletal symptoms. 76 Many of the risk factors for chronicity identified in this review fall under this category. The interest in yellow flags originates from the concept that early interventions might avert the development of disability. When patient selection is performed accurately and when an intervention known to address these factors is competently applied, good outcomes are to be expected. 96

Limitations of this review
A major limitation of this review was that only 1 high-quality study was detected in our literature search. Loss to follow-up was significant in many fair-quality studies, and this reduced the   (continued on next page) 6 (2021) e919 www.painreportsonline.com number of good-quality studies. Furthermore, chronic low back pain as an outcome is hard to validate since it is always more or less self-reported. Many studies have tried to minimize this bias by using validated questionnaires. Nine of the studies (36%) used the same population data from HUNT studies. 46,[57][58][59][60][61][62][63]97 The results that were only observed from HUNT studies were body height 60 and measures, 59 diabetes, 57 blood pressure, 46,62 and pulse pressure. 62 However, the risk of bias in this particular study population can be assessed as low because of the large sample size and long follow-up period. The Nord-Trondelag Health Studies (HUNT studies) were population-based health surveys conducted in 1984 to 1986, 1995 to 1997, and 2006 to 2008. All residents older than 20 years of the entire Norwegian county were invited to take part in these large surveys. 63 Some risk factors that seemed similar and were detected in multiple studies differed nonetheless to some extent in definition or measurement choice. To avoid too much heterogeneity inside 1 risk factor, they were intentionally not combined. Thus, it was difficult to reach a strong conclusion about the significance of several risk factors because they were only evaluated by a small number of studies.
Defining CLBP as persistent pain for at least 3 months is an artificial means of controlling the heterogenic population with LBP symptoms. Evidence from long-term studies indicates that people with long-term problems can have pain episodes separated by periods that are pain free, periods of continuous mild pain with low impact, or periods of severe pain with a large impact on their lives. 25 When finding a potential association between a prognostic factor and an outcome, one must not assume that the effect is direct and isolated. Nonspecific low back pain is a multifactorial and complex condition with the impact of different factors changing over time. 32 This review simply identifies the factors related to chronicity; it does not, however, study whether the presence of 1 factor is sufficient or whether a certain mix of factors is required. Therefore, when developing more comprehensive models that include connections between these factors, it is essential to consider which factors are truly important.

Usefulness of results and recommendations
A "wait and see" approach is no longer advisable because early screening provides reliable and valuable information for identifying those at risk of delayed recovery and for formulating a treatment strategy from the start. 81 The subgrouping of patients with nonspecific LBP and finding tailored treatments and management strategies are the main research priorities in the field of LBP. 16 It is therefore important to detect those patients at risk of developing chronicity in the early phases of the symptoms and to offer tailored treatment according to the risks in question. Especially stratification according to psychosocial risk factors has achieved promising results, 34,65 but the disadvantage is the lack of work-related items, socioeconomic variables, and symptom factors. Then, additional steps may be needed to identify the specific problems of patients to improve outcomes. 81 The findings of this review may be helpful in the planning of future studies concerning the prevention of CLBP and to aid clinicians detect patients at risk of chronicity.

Disclosures
The authors have no conflicts of interest to declare.