UNIVERSIDADE FEDERAL DE MINAS GERAIS Escola de Educação Física, Fisioterapia e Terapia Ocupacional Programa de Pós-Graduação em Ciências da Reabilitação Samuel Silva ASSOCIAÇÃO ENTRE SONO E DESFECHOS CLÍNICOS EM INDIVÍDUOS COM DOR LOMBAR Belo Horizonte 2023 Samuel Silva ASSOCIAÇÃO ENTRE SONO E DESFECHOS CLÍNICOS EM INDIVÍDUOS COM DOR LOMBAR Dissertação apresentada ao curso de Mestrado em Ciências da Reabilitação da Escola de Educação Física, Fisioterapia e Terapia Ocupacional da Universidade Federal de Minas Gerais, como requisito parcial à obtenção do título de Mestre em Ciências da Reabilitação. Orientadora: Profa. Dra. Andressa Silva Coorientadores: Prof. Dr. Rafael Zambelli Pinto e Profa. Dra. Jill Hayden Área de Concentração: Desempenho funcional humano Belo Horizonte 2023 S586a 2023 Silva, Samuel Associação entre sono e desfechos clínicos em indivíduos com dor lombar. [manuscrito] / Samuel Silva – 2023. 144 f.: il. Orientadora: Andressa Silva Coorientador: Rafael Zambelli Pinto Coorientadora: Jill Hayden Dissertação (mestrado) – Universidade Federal de Minas Gerais, Escola de Educação Física, Fisioterapia e Terapia Ocupacional. Bibliografia: f. 120-123 1. Idosos – Saúde e higiene – Teses. 2. Sono – Teses. 3. Transtornos do sono – Teses. 4. Dor lombar – Teses. I. Silva, Andressa. II. Pinto, Rafael Zambelli. III. Hayden, Jill. IV. Universidade Federal de Minas Gerais. Escola de Educação Física, Fisioterapia e Terapia Ocupacional. V. Título. CDU: 615.8 Ficha catalográfica elaborada pela bibliotecária Sheila Margareth Teixeira Adão, CRB 6: n° 2106, da Biblioteca da Escola de Educação Física, Fisioterapia e Terapia Ocupacional da UFMG. “Eu quero ser maior que essas muralhas que eles construíram ao meu redor” (ABEBE BIKILA, 2018) AGRADECIMENTOS (ACKNOWLEDGEMENTS) Uma longa jornada se encerra aqui. Com certeza, nada seria possível sem a ajuda de tantas pessoas maravilhosas que passaram pelo meu caminho durante estes dois anos. Durante meu mestrado, pude entender o verdadeiro significado da expressão “enjoy the process”. Nunca tive pressa de encerrar pois, para mim, tudo nesses dois anos foi muito prazeroso. Com certeza, foi o melhor período da minha vida! Obrigado a todos e todas que, de alguma forma, acreditaram em mim e me apoiaram durante esse percurso! Primeiramente, gostaria de agradecer a Deus por ter me dado a força e coragem necessária para passar por esse processo. Um agradecimento especial aos meus pais, Maria Antonia e Waldeir, que sempre foram minha base, pavimentaram o caminho para que eu pudesse estar aqui e me apoiam em todas as minhas decisões. Tudo isso é por vocês! Obrigado também aos meus avós, Darcisa, Joaquim (in memorian), Divina e Juca, meus tios e tias, Vinícius, Sônia, Paulo, Ana, Varlei, Mauro, Cidinha, Rosana e Graça, ao meu primo Marcos, e ao meu padrinho e madrinha Joel e Neuza por serem meus pilares. Sem dúvidas, sem minha orientadora, Profa. Andressa, nada disso teria acontecido. Obrigado por, junto ao Prof. Marco Túlio, ter aberto as portas do CEPE para mim, por confiar em mim e por ter me despertado o desejo de estar envolvido na construção do conhecimento científico. Serei eternamente grato por tudo! Muita gratidão também ao meu coorientador Prof. Rafael Zambelli. Obrigado por me ensinar tanto durante esse período e por ser uma grande referência e inspiração para mim! Ainda, obrigado a todos os meus colegas de grupo de pesquisa, integrantes do CEPE e do grupo Coluna Saudável 60+, que estiveram ao meu lado nessa caminhada compartilhando os “perrengues” e me ensinaram tanto, em especial, os(as) queridos(as) Gabriel, Caique, Eduardo, Isadora, Gerônimo, Renato e Eleonora. Um agradecimento também aos alunos de iniciação científica Vitória e Raimundo que foram essenciais na condução dos estudos. É impossível expressar em palavras o tamanho da minha gratidão por tudo que a Profa. Jill Hayden fez e tem feito por mim. Muito obrigado por ter aberto as portas do seu grupo de pesquisa para mim, por ter me ensinado tanto (e continuar me ensinando), por ser tão atenciosa, generosa, gentil e um exemplo de ser humano. Um agradecimento especial também a Rachel por todo o suporte e carinho durante meus 6 meses em Halifax. Estendo meu agradecimento a todos os membros do BACK Program team, aos professores e funcionários do Department of Community Health and Epidemiology da Dalhousie Univerisity por terem me recebido de forma tão calorosa. Essa experiência foi, sem dúvidas, um marco na minha vida! E claro, um agradecimento aos meus amigos Pedro, Bruno e Leo por tudo que compartilhamos, dos momentos felizes (que foram, sem dúvidas, a imensa maioria) até as dificuldades de estar longe de casa e da família. Muito obrigado a todos os meus amigos, em especial, ao Álamo, Marcos, Lucas Nunes, Lucas Martins, Eduardo, Thais, Yuri, Dênis, Matheus Pessi, Mateus Batista, Alisson, Rayane e Estevão por estarem sempre me apoiando, dando força e demonstrando sempre estarem felizes por minhas conquistas. Tenho muito orgulho de onde eu vim, e por isso, me sinto um representante do bairro onde cresci. Tudo que aprendi, com todos que, desde a minha infância, tiveram ao meu redor e me ensinaram tanto no bairro Cidade Jardim (Três Pontas/MG), se reflete um pouco aqui. Não poderia deixar de agradecer o meu professor de Jiu-Jitsu, Jô, que lá atrás, me incentivou a cursar fisioterapia, e ouvindo-o, descobri uma paixão. Mais um agradecimento super especial, agora, a minha companheira, Aline. Obrigado por ter sido meu alicerce durante esse período, por estar comigo nas dificuldades e ter deixado tudo mais leve e prazeroso. Te amo muito! Finalmente, e não menos importante, gostaria de agradecer a todos os funcionários da EEFFTO, em especial a Eliane, por serem sempre tão solícitos e gentis. Obrigado a todos os professores da escola que me ensinaram tanto durante as disciplinas e obrigado as agências de fomento FAPEMIG e Global Affairs Canada que permitiram com que eu conseguisse dedicar o tempo necessário aos trabalhos conduzidos no meu mestrado. RESUMO OBJETIVOS: O objetivo do estudo um foi de revisar a literatura sistematicamente e investigar se o sono se associa com desfechos clínicos futuros em adultos com dor lombar (DL). Os objetivos do estudo dois foram i) investigar a associação da quantidade e eficiência de sono medidas objetivamente com mudanças em desfechos clínicos em idosos com DL crônica que receberam tratamento fisioterapêutico; e ii) examinar a associação transversal da quantidade, eficiência, latência, e fragmentação de sono com a catatrofização da dor. MÉTODOS: O estudo um foi uma revisão sistemática com meta-análises de estudos de coorte prospectivos e análises secundárias de ensaios clínicos aleatorizados. O estudo dois foi um estudo de coorte prospectivo com seguimento de dois meses que incluiu idosos (≥60 anos) com DL crônica que estavam iniciando tratamento fisioterapêutico no local de recrutamento. RESULTADOS: O estudo um incluiu 14 estudos, totalizando 19.170 participantes. Treze estudos foram classificados com alto risco de viés. Com base em uma abordagem de vote-counting, foram encontradas associações entre sono na linha de base e intensidade da dor futura e recuperação da DL; e entre mudanças no sono e mudanças na intensidade da dor, mudanças na incapacidade e recuperação da DL. Baixa qualidade de sono na linha de base foi associada moderadamente com a não melhora geral da DL no longo-muito longo prazo (OR=1,55; IC 95% 1,39 a 1,73; três estudos fornecendo tamanhos de efeito não ajustados), e a não melhora do sono foi associada fortemente com a não melhora geral da DL no curto-médio prazo (OR=3,45; IC 95% 2,54 a 4,69; quatro estudos fornecendo tamanhos de efeito não ajustados). Não foram encontradas associações entre sono na linha de base e incapacidade futura e melhora geral da DL no curto-médio prazo. Todos os achados foram sustentados por uma baixa-muito baixa qualidade de evidência. O estudo dois incluiu 51 participantes com seguimento completo (60,8% mulheres; idade média de 70,1±5,6 anos). Não foram encontradas associações entre qualidade e eficiência de sono e mudanças na intensidade da dor, mudanças na incapacidade e recuperação autorrelatada da DL na avaliação de seguimento. Uma correlação postiva foi encontrada entre fragmentação de sono e catastrofização da dor (r=0,30; IC 95% 0,03 a 0,54), no entanto, a associação não foi encontrada após o ajuste por potenciais confundidores. CONCLUSÕES: Nossos resultados do estudo um indicaram que o sono autorrelatado parece se associar com desfechos futuros de DL e a mudanças no sono parecem se associar com mudanças na DL. Com base nos resultados do estudo dois, a quantidade e eficiência de sono mensuradas objetivamente parecem não se associar com mudanças nos desfechos de DL após tratamento fisioterapêutico em idosos com dor lombar crônica. A fragmentação do sono mensurada objetivamente parece ser o domínio do sono com a relação mais forte com catastrofização da dor. Palavras-chave: Transtornos do Sono do Ritmo Circadiano. Actigrafia. Dor lombar. Não específica. Dor crônica. Idoso. Revisão sistemática. Prognóstico. ABSTRACT OBJECTIVES: The objective in study 1 was to systematically review the literature investigating whether sleep is associated with future clinical outcomes in adults with low back pain (LBP). The objectives in study 2 were i) to investigate the association between objectively measured sleep quantity and efficiency with changes in clinical outcomes in older adults with chronic LBP receiving physical therapy care; and ii) to examine the cross-sectional association between objectively measured sleep quantity, efficiency, onset latency, and fragmentation with pain catastrophizing. METHODS: Study 1 was a systematic review with meta-analyses of prospective cohort studies and secondary analyses of randomized controlled trials. Study 2 was a prospective cohort study with a 2-month follow-up that included older adults (≥60 years old) with chronic LBP initiating physical therapy care at the recruitment setting. RESULTS: Study 1 included 14 studies, totaling 19,170 participants. Thirteen studies were rated as having high risk of bias. Based on a vote-counting approach, associations were found between baseline sleep with future pain intensity, LBP recovery, and between changes in sleep with changes in pain intensity, changes in disability, and LBP recovery. Baseline poor sleep was moderately associated with non-improvement in LBP in the long-very long term (OR=1.55, 95%CI 1.39 to 1.73; three studies providing unadjusted effect sizes), and non-improvement in sleep was largely associated with non-improvement in LBP outcomes in the short-moderate term (OR=3.45, 95%CI 2.54 to 4.69; four studies providing unadjusted effect sizes). No association was found between baseline sleep with future disability and overall LBP improvement in the short-moderate term. All findings were supported by low to very low-quality of evidence. Study 2 included 51 participants with complete follow-up assessments (60.8% women; mean age 70.1±5.6 years). No association was found between sleep quantity and sleep efficiency with changes in pain intensity, changes in disability, and self-reported recovery at follow-up. A positive correlation was found between sleep fragmentation and pain catastrophizing (r=0.30, 95%CI: 0.03 a 0.54); however, no association was found when adjusting for potential confounders. CONCLUSIONS: Our results from study 1 indicated that self-reported sleep seems to be associated with future LBP outcomes and changes in sleep seem to be associated with changes in LBP. Based on the results from study 2, objectively measured sleep quantity and sleep efficiency may not be associated with changes in LBP outcomes after physical therapy care in older adults with chronic LBP. Moreover, objectively measured sleep fragmentation seems to be the sleep domain with the strongest relationship with pain catastrophizing. Keywords: Sleep arousal disorders. Actigraphy. Low back pain. Nonspecific. Chronic pain. Aged. Systematic review. Prognosis. LIST OF TABLES AND FIGURES STUDY 1 Figure 1. Framework for the potential confounders of the association between sleep and low back pain outcomes. Predefined potential confounders were age, psychological/occupational factors (e.g., anxiety, depression, catastrophizing, job satisfaction, work status), smoking habits, body mass index, general health (e.g., physical activity level, comorbidities), and clinical low back pain characteristics (e.g., baseline pain intensity, baseline disability, low back pain duration). Figure created by the authors.………………………...……………………………………..…………………28 Figure 2. Flowchart of the review selection process………….………………………..32 Table 1. Characteristics of the included studies. …….………………………………...34 Figure 3. Graphs illustrating our vote-counting approach with the number of studies, their respective sample sizes, and reported associations (positive, no association, or negative) for baseline sleep and outcomes: a. future pain intensity, b. disability, and c. recovery. Each bar represents a sleep domain evaluated by an individual study; the bar height represents the study sample size. Bars in black represent a ‘positive association’ and gray bars represent ‘no association’. No study found a negative association. *=Studies that evaluated two sleep domains are represented twice………………………………………………………………………………………….39 Figure 4. Forest plot of the unadjusted (4a) and adjusted (4b) associations between baseline sleep and chance of non-improvement in short-moderate term (3 to 6 months of follow-up)…………………..………………………………..…………………40 Figure 5. Forest plot of the unadjusted (5a) and adjusted (5b) associations between baseline sleep and chance (5a)/ risk (5b) of non-improvement in long-very long term (≥12 months of follow-up).……………………………………………….……….………41 Figure 6. Forest plot of the unadjusted association between changes in sleep and chance of non-improvement in low back pain outcomes in short-moderate term (3 to 6 months of follow-up). ……………………………………………………..….…………42 Supplementary Table 1. Reported and calculated effect sizes of included studies…………………………………………………………………….…………………63 Supplementary Table 2. Risk of bias assessment using the Quality in Prognosis Studies (QUIPS) tool………………………….……………………………………………67 Supplementary Table 3. Grading of recommendations assessment, development and evaluation (GRADE) judgements for the available evidence……………………..68 STUDY 2 Figure 1. Flowchart of the selection process……………………………………………95 Table 1. Baseline sociodemographic, sleep, and clinical characteristics…………….96 Table 2. Unadjusted and adjusted coefficients from the simple and multivariable associations between total sleep time and sleep efficiency at baseline as independent variables with changes in pain intensity after the 8-week follow-up as the dependent variable……………………………………...………………………….…99 Table 3. Unadjusted and adjusted coefficients from the simple and multivariable associations between total sleep time and sleep efficiency at baseline as independent variables with changes in disability after the 8-week follow-up as the dependent variable……………………………………………………………………....100 Table 4. Unadjusted and adjusted coefficients from the simple and multivariable associations between total sleep time and sleep efficiency at baseline as independent variables with self-perceived recovery after the 8-week follow-up as the dependent variable……………………………………………………………………..…101 LIST OF ABBREVIATIONS AND ACRONYMS ALBP acute low back pain BMI body mass index CI confidence interval CLBP chronic low back pain GDS-15 Getriatric Depression Scale GPE Global Perceived Effect GRADE Grading of Recommendations Assessment, Development and Evaluation IQR interquartile range LBP low back pain NREM sleep non-rapid eye movement sleep NRS Numerical Rating Scale OR odds ratio PCS Pain Catastrophizing Scale PRISMA Preferred Reporting Items for Systematic Reviews and Meta- Analysis guidelines PSQI Pittsburgh Sleep Quality Index QUIPS Quality In Prognosis Studies REM sleep rapid eye movement sleep RMDQ Roland-Morris Disability Questionnaire RMQ Roland-Morris Questionnaire RR risk ratio SD standard deviation STROBE Strengthening the Reporting of Observational Studies in Epidemiology TABLE OF CONTENTS 1. INTRODUCTION ................................................................................................... 17 2. STUDY 1 ............................................................................................................... 21 3. STUDY 2 ............................................................................................................... 82 4. FINAL CONSIDERATIONS ................................................................................ 118 REFERENCES ........................................................................................................ 120 APPENDICES ......................................................................................................... 124 Appendix A – Informed consent form ...................................................................... 124 Appendix B – Evaluation form ................................................................................. 128 ANNEXES ............................................................................................................... 130 ANNEX 1 – Ethics committee approval letter .......................................................... 130 ANNEX 2 – Sleep log .............................................................................................. 135 ANNEX 3 – Geriatric Depression Scale .................................................................. 137 ANNEX 4 – Numerical Rating Scale ........................................................................ 138 ANNEX 5 – Roland-Morris Disability Questionnaire ................................................ 139 ANNEX 6 – Global Perceived Effect Scale .............................................................. 140 ANNEX 7 – Pain Catastrophizing Scale .................................................................. 141 MINI RESUME ........................................................................................................ 142 PREFACE This thesis, entitled “Association between sleep and clinical outcomes in individuals with low back pain” follows the criteria established by the Graduate Program in Rehabilitation Sciences and is formatted based on the standards of the Associação Brasileira de Normas Técnicas (ABNT). Two studies were conducted for the development of this thesis. Study 1 is a systematic review entitled “Sleep as a prognostic factor in low back pain: a systematic review with meta-analyses of prospective cohort studies and secondary analyses of randomized controlled trials”. Study 2, entitled “Association between objectively measured sleep and clinical outcomes in older adults with chronic low back pain receiving physical therapy care: a prospective cohort study”, is a prospective cohort study that was pre-planned and designed in an attempt to fill some of the gaps in the literature, highlighted in study 1. Firstly, this thesis presents a broad introduction to contextualize the topic addressed. Secondly, the two studies are presented in the same format in which they were submitted to the respective journals, following journal standards (including all data submitted as supplemental materials and appendices). Study 1 is under review by the PAIN Journal and the revised version of study 2 is under review by the European Journal of Pain. After the presentation of the studies, there is a section for final considerations where we intended to interpret and summarize the findings of both studies and discuss potential scientific and clinical implications of these findings. Finally, we describe the references, cited in the introduction section; appendices, from study 2; annexes, also from study 2; and a mini resume as required by the graduate program. 17 1. INTRODUCTION Low back pain (LBP) has been defined as pain or discomfort located between the last rib and above the inferior gluteal fold, with or without referred pain to the leg (Collaborators, 2023). LBP is classified based on its etiology into specific and non- specific. LBP is considered specific when there is a clear and recognizable cause for the pain symptoms (e.g., fracture, tumor, radiculopathy), and LBP is considered non- specific when the underlying causes are not clearly identifiable (Balagué et al., 2012). Non-specific LBP accounts for around 90% of all LBP cases (Maher; Underwood; Buchbinder, 2017). LBP is further categorized into acute and chronic according to its persistence. Chronic LBP stands for LBP lasting for 12 weeks or more, subacute LBP stands for LBP symptoms present for 6 weeks to less than 12 weeks, and acute LBP is when symptoms are present for less than 6 weeks (Deyo et al., 2014). Most acute LBP episodes have a positive prognosis with resolution of symptoms within 12 weeks (Chou; Shekelle, 2010); however, when LBP becomes chronic, it can represent a major burden on healthcare systems worldwide (Chou; Shekelle, 2010). In 2018, a call for action paper was published by The Lancet alerting to the need to prioritize LBP as a public health problem globally (Buchbinder et al., 2018). It is estimated that 70-80% of the adult population will experience LBP at least once in their lifetime (RUBIN, 2007). Evidence suggests that the prevalence of LBP in adults has been increasing over the past three decades, and some recent estimates point to a continuous increase in the next decades (Collaborators, 2023; Wu et al., 2020). Due to its high prevalence and its potential to cause severe disability, LBP results in tremendous societal cost, for healthcare systems, patients, and employers (e.g., absenteeism/presenteeism) (Coombs et al., 2021; Dieleman et al., 2020; Van der Wurf et al., 2021). Along with neck pain, LBP is the leading cause of years lived with disability in low-, mid- and high-income countries (Chen et al., 2021). There is a clear need to understand factors that may be associated with poor outcomes and chronicity of LBP. Musculoskeletal pain conditions have been recognized by the literature as complex conditions that require multidimensional management approaches that incorporate biopsychological aspects (Cholewicki et al., 2019). For instance, there is compelling evidence that prognostic factors in musculoskeletal pain conditions are 18 multidimensional (Artus et al., 2017; Nieminen; Pyysalo; Kankaanpää, 2021). The definition of pain from the International Association for the Study of Pain states that pain is defined as “an unpleasant sensory and emotional experience associated with, or resembling that associated with, actual or potential tissue damage”, reinforcing that pain is a subjective experience rather than a true reflection of tissue state (Raja et al., 2020). Healthcare providers and clinical researchers need to shift from a biomedical framework to a biopsychosocial framework of care/research in LBP management (Buchbinder et al., 2018; O’Sullivan, 2011). In this sense, other aspects of life, such as sleep, may be important for understanding the processes and prognosis of musculoskeletal pain. Sleep is a fundamental physiologic process for humans and is a biological requirement for life (Grandner, 2016). Human sleep is divided into two major phases: non-rapid eye movement sleep (NREM sleep) and rapid eye movement sleep (REM sleep). The former is further divided into three phases: N1, N2, and N3 (also known as slow-wave sleep). REM sleep is often related to cognitive and mental recovery (Peever; Fuller, 2016). Li et al. (2017) showed that REM sleep has a role in maintaining new synapses after motor learning (Li; Vitiello; Gooneratne, 2017). NREM sleep is primarily associated with metabolic and physical recovery. In the N3 phase, for example, the growth hormone secretion reaches its peak (Cauter; Copinschp, 2000). REM sleep and NREM sleep repeat themselves in 90-minutes cycles for about 4 to 6 times per night. Each phase begins with lighter NREM sleep (i.e., N1 and N2 phases), followed by deeper NREM sleep (i.e., N3 phase), and then REM sleep. Typically, in healthy adults, 50% of night sleep is composed of N1 and N2 phases, 20% of N3 phase, 25% of REM sleep, and 5% of awake periods (Copinschi; Caufriez, 2013). However, the human sleep pattern changes throughout the lifetime. With aging, there is a decrease in total sleep time, slow-wave sleep, REM sleep, and sleep efficiency, associated with an increase in sleep onset latency, awakenings after sleep onset, and duration of lighter sleep phases (i.e., N1 and N3 phases of NREM sleep) (Moraes et al., 2014; Ohayon et al., 2004). Although it is widely known that lack of sleep is associated with several poor health outcomes such as cardiovascular, neurological, and chronic pain conditions, in today’s modern society, sleep has become a low-priority component in humans lives (Coveney, 2014; Liew; Aung, 2021; Ohara; Honda; Hata, 2018; Silva et al., 2022; 19 Uhlig et al., 2018; Yin et al., 2017). There has been an increase in demand and pressure for productivity, which may lead to depreciation of rest periods and reduced bedtime (Coveney, 2014). In addition, the increasing and excessive use of smartphones and other electronic devices, especially during nighttime, has contributed to changes in the sleep pattern of the modern society (Sohn et al., 2021). The World Health Organization stated that there is an existing public health epidemic of sleepiness due to lack of sleep (Lyon, 2019). A previous study showed that it appears that humans are sleeping about 6 minutes less each decade (Kronholm et al., 2008). There is robust evidence supporting the bidirectionality of the pain-sleep relationship, where pain symptoms tend to impair sleep and poor/lack of sleep may increase and facilitate pain (Azevedo et al., 2011; Finan; Goodin; Smith, 2013). However, studies comparing how one variable affects the other have shown that sleep seems to have a greater influence on pain than the opposite (Finan; Goodin; Smith, 2013; Morelhão et al., 2022). Sleep problems are very common in people who live with LBP. A recent systematic review found that around 72% of individuals with chronic back pain have poor sleep quality, compared with 23% of pain-free individuals (Sun et al., 2021). A previous overview reported that individuals with pain conditions tend to have shorter sleep duration, more fragmented sleep, longer sleep onset latency, less sleep efficiency, shorter REM sleep and deeper sleep (i.e., phase N3 of NREM sleep), and longer lighter sleep (i.e., phases N1 and N2 of NREM sleep) (Lavigne et al., 2011). Sleep restriction might dysregulate endogenous opioid pathways, which are involved in the descending inhibitory system (Nijs et al., 2018). This can lead to an impaired control of nociceptive inputs, which can further lead to increased pain sensitization and decreased pain habituation, facilitating hyperalgesia (Finan; Goodin; Smith, 2013; Nijs et al., 2018; Silva et al., 2018; Simpson et al., 2018). Dopaminergic and serotoninergic pathways are involved in modulating the sleep- awake cycle and pain perception; therefore, it has been proposed that impairment in these pathways may partially explain how sleep restriction might contribute to exacerbating pain (Finan; Goodin; Smith, 2013; Nijs et al., 2018). Moreover, sleep restriction stimulates the release of pro-inflammatory cytokines, which are potential nociceptive inputs, and have been associated with pain chronicity (Grandner, 2016; 20 Nijs et al., 2018; Roehrs; Roth, 2005). Finally, sleep might also be associated with the way symptoms are perceived by the individual with pain. Sleep restriction and poor sleep may promote a state of anxiety and hypervigilance (Nijs et al., 2018). Motomura et al. (2017) showed that sleep deprivation can decrease the connectivity between the amygdala and the medial prefrontal cortex, which can decline mood and affect emotions (Motomura et al., 2017). This may be associated with increased irritability and ruminative thinking, which can lead to increased catastrophizing behavior toward pain symptoms (Gerhart et al., 2016; Whibley et al., 2019). Considering the potential influence of sleep on pain processing and perception as presented above, it is relevant to investigate the prognostic value of sleep in LBP, understanding how sleep may be associated with future clinical outcomes in this population. A prognostic factor is a variable associated with a subsequent health outcome among people with a given health condition (Riley et al., 2013, 2019). Prognostic factor studies are one of four categories of prognostic research (i.e., fundamental prognosis, prognostic factor, prognostic model, and stratified medicine) (Hemingway et al., 2013). Prognostic factor research is further subcategorized into exploratory (i.e., investigating the role of multiple potential prognostic factors) and confirmatory (i.e., investigating the role of a single prognostic factor) studies (Riley et al., 2013). Therefore, our objective with this thesis was to comprehensively investigate the role of sleep as a prognostic factor in LBP and fill some of the gaps in the literature by conducting a primary study. 21 2. STUDY 1 Submitted to the SLEEP Journal (https://academic.oup.com/sleep) Title: Sleep as a prognostic factor in low back pain: a systematic review with meta-analyses of prospective cohort studies and secondary analyses of randomized controlled trials Authors: Samuel Silvaa,b, Jill Alison Haydenb, Gabriel Mendesa, Arianne Verhagenc, Rafael Zambelli Pintoa, Andressa Silvaa Affiliations: aUniversidade Federal de Minas Gerais, Belo Horizonte, Brazil bDepartment of Community Health and Epidemiology, Dalhousie University, Halifax, Canada cGraduate School of Health, University of Technology Sydney, Sydney, Australia Corresponding author: Andressa Silva, PhD School of Physical Education, Physical Therapy, and Occupational Therapy Universidade Federal de Minas Gerais Av. Antônio Carlos, 6627, Pampulha, CEP: 31270-901 Belo Horizonte, MG, Brazil. E-mail: andressa@demello.net.br. Phone number: +55 (31) 2513-2347. 22 ABSTRACT Sleep problems are common in individuals with low back pain (LBP) and sleep restriction seems to be associated with impaired pain processing. Our objective was to investigate whether sleep is associated with future outcomes in adults with LBP. We conducted a systematic review with meta-analyses of prospective cohort studies and secondary analyses of randomized controlled trials (registration - PROSPERO CRD42022370781). In December 2022, we searched the MEDLINE, Embase, CINAHL, and PsycINFO databases. Fourteen studies, totaling 19,170 participants were included. Thirteen studies were rated as having high risk of bias (QUIPS tool). Based on a vote-counting approach, we found associations between baseline sleep with future pain intensity, recovery, and between changes in sleep with changes in pain intensity, changes in disability, and recovery. We further synthesized outcomes as ‘overall LBP improvement’ outcome and sleep domains as ‘good sleep’ versus ‘poor sleep’ or ‘improvement in sleep’ versus ‘non-improvement in sleep’ exposures. Baseline poor sleep was moderately associated with non-improvement in LBP in the long-very long term (OR 1.55, 95% CI 1.39 to 1.73; three studies providing unadjusted effect sizes), and non-improvement in sleep was largely associated with non-improvement in LBP outcomes in the short-moderate term (OR 3.45, 95% CI 2.54 to 4.69; four studies providing unadjusted effect sizes). We found no association between baseline sleep with future disability and overall LBP improvement in the short-moderate term. All findings were supported by low to very low-quality of evidence. Future high-quality primary studies are needed to strengthen our certainty about the evidence. KEY WORDS: Low Back Pain, Chronic Pain, Sleep Arousal Disorders, Prognosis, Systematic Review. 23 INTRODUCTION It is estimated that 70-80% of the adult population will experience low back pain (LBP) at least once in their lifetime 1. Evidence suggests that the prevalence of LBP in adults has been increasing over the past three decades, and some recent projections point to a continuous increase in the next decades 2,3. Due to its high prevalence and its potential to cause severe disability, LBP results in tremendous societal cost, for healthcare systems, patients, and employers (e.g., absenteeism/presenteeism) and is the leading cause of years lived with disability in low-, mid- and high-income countries 4–7. There is a clear need to understand factors that may be associated with poor outcomes and chronicity of LBP. Sleep problems are very common in people who live with LBP. A recent systematic review found that 72% of individuals with chronic back pain have poor sleep quality, compared with 23% of pain-free individuals 8. In addition, a previous overview reported that individuals with musculoskeletal pain conditions tend to have shorter sleep duration, more fragmented sleep, longer sleep onset latency, and less sleep efficiency 9. Furthermore, previous studies have found a decreased pain threshold and less pain habituation in individuals with sleep restriction 10–13. Sleep restriction can affect the descending pain modulatory system due to the impairment of endogenous opioid systems and serotonergic and dopaminergic pathways 14, in addition to increasing inflammatory cytokine levels which have been associated with pain chronicity 15,16. Experts in the field have stated that clinicians should assess sleep in individuals seeking treatment for LBP, as sleep disturbances are potentially associated with worse LBP outcomes 17. However, findings from prospective cohort studies are inconsistent 18,19, and as far as we know, no review has comprehensively investigated whether sleep is associated with future LBP outcomes. Therefore, our aim was to systematically review the literature and investigate whether sleep is associated with future outcomes (i.e., pain intensity, disability, and recovery) in adults with LBP. MATERIAL AND METHODS 24 We conducted a systematic review of prospective cohort studies. The protocol was prospectively registered on PROSPERO (CRD42022370781). We have reported this review following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines (PRISMA) 20. Search strategy We conducted searches of electronic databases using free text terms and subject headings related to LBP, sleep, and cohort/prognostic studies (inception to December 2022): MEDLINE via Ovid, Embase (www.embase.com), CINAHL via EBSCO, and PsycINFO via EBSCO (Appendix A). We supplemented our electronic search by: 1. hand searching of the reference lists of broad systematic reviews investigating prognostic factors in LBP and reviews on the relationship between sleep and LBP, 2. searching the reference lists of all included studies, and 3. citation searching the primary publications of the Pittsburgh Sleep Quality Index (PSQI) (the most common sleep measurement tool used in the field) 8,21. Study selection criteria Population We included studies if 75% or more of the sample was aged over 18 years; had non-specific LBP (pain or discomfort located between the last rib and above the inferior gluteal fold, with or without referred pain to the leg 3), regardless of the duration of symptoms. Studies that mixed non-specific LBP with specific LBP (e.g., stenosis, spondylolisthesis, disc herniation confirmed by image screening, pregnancy-related, LBP after back surgery), with other pain conditions, or with healthy individuals were excluded unless ≥75% of the sample had non-specific LBP or if effect sizes could be extracted separately for the subgroup with non-specific LBP. Prognostic factors 25 We included studies that evaluated at least one sleep domain at baseline, regardless of the measures used. However, we predefined which measures would be considered valid for each variable to inform our risk of bias assessment and sensitivity analyses: 1. Sleep quality defined according to Kline (2013) as the individual’s self-satisfaction with the sleep experience 22. We considered the PSQI 21 as valid and reliable measure for self-reported sleep quality. 2. Sleep quantity defined as the total time a person actually spends sleeping 23. We considered objective sleep measures (i.e., actigraphy and polysomnography) as valid and reliable measures of sleep quantity 24. 3. General insomnia symptoms characterized by difficulties in initiating and maintaining sleep 25. Standardized scales and questionnaires, including the Insomnia Severity Index 25, and the Athens Insomnia Scale 26 were considered valid tools for measuring general insomnia symptoms. 4. Daytime sleepiness defined as “daily episodes of an irrepressible need to sleep or daytime lapses into sleep” 27. Standardized scales and questionnaires including the Epworth Sleepiness Scale 28 and the Karolinska Sleepiness Scale 29 were considered valid tools for measuring daytime sleepiness. 5. Sleep efficiency defined as the total sleep time divided by time in bed 23. We considered objective sleep measures (i.e., actigraphy and polysomnography) as the valid and reliable measures of sleep efficiency 30. 6. Sleep fragmentation defined as the measure of the number of awakenings and/or time awake after sleep onset. We considered objective sleep measures (i.e., actigraphy and polysomnography) as the valid and reliable measures of these variables 30. 26 7. Sleep onset latency defined as the time one takes to fall asleep after going to bed 23. We considered polysomnography as the valid and reliable measure of sleep onset latency 31. Outcomes We included studies that evaluated at least one of our outcomes of interest: pain intensity, disability, and recovery of LBP. For pain intensity, we included studies that used the Visual Analogue Scale (VAS), Numerical Rating Scale (NRS), or the McGill Pain Score 32. For disability, we included studies that used tools designed to measure LBP-related functional limitations such as the Roland-Morris Questionnaire (RMQ) 33 and the Oswestry Disability Index 34. For recovery of LBP, we included studies that measured self-perceived recovery scales such as the Global Rating of Change Scale 35, and Global Perceived Effect Scale 36. Studies that dichotomized the outcome as presence/absence of LBP at follow-up using simple questions or screening tools such as the Nordic Musculoskeletal Questionnaire 37 were also included. Studies that used measures of pain intensity or disability and dichotomized the outcome (i.e., as having/not having pain or disability at follow-up) were considered as reporting a recovery outcome. Measures of self-perceived recovery were prioritized in our data synthesis when multiple measures of recovery were available. Study design We included prospective cohort studies and secondary analyses of randomized controlled trials (any language of publication) with follow-up of ≥ 3 months that reported the association (simple or multivariable) between at least one sleep domain and one of our outcomes of interest. In cases of multiple studies using overlapping data, we considered the study with the largest sample size as the primary report. For linked publications providing different useful data (e.g., different outcomes), we considered the publications as one study and the first one published was defined as the primary report. Study selection 27 Two independent reviewers (SS, GM) conducted title and abstract screening, then full text review using a web-based systematic review platform, Covidence (www.covidence.org). In cases of disagreement after discussion, a third reviewer (JAH) was consulted to arbitrate. Data extraction Two independent reviewers (SS, GM) performed data extraction using Covidence. Based on the recommendations of the CHARMS-PF checklist 38, we extracted the following data: study design, country of conduct, recruitment setting, phase of investigation, study conduct dates, baseline sample characteristics, sample size, follow-up duration, sleep measures, outcome measures, effect sizes, and covariates adjusted in the statistical analysis. If any essential information, such as sample size, sample characteristics or any relevant statistical data was unclear, the corresponding author was contacted via e-mail. In cases of no response, we considered the data as unclear or missing. Risk of bias assessment Two independent reviewers (SS, GM) assessed risk of bias with a third reviewer (JAH) arbitrating in cases of disagreement. We used the Quality In Prognosis Studies (QUIPS) tool 39, evaluating 6 bias domains: study participation, study attrition, prognostic factor measurement, outcome measurement, study confounding, and statistical analysis and reporting (Appendix B). The assessors rated each domain as having high, moderate, or low risk of bias. We rated the overall risk of bias in each study as low (low risk of bias in all domains), some concerns (moderate and low risk of bias in all domains), and high risk of bias (high risk of bias in at least one domain). Potential confounders Based on the current literature 40–49, we predetermined potential confounders of the relationship between sleep and LBP outcomes (i.e., variables potentially associated with both exposure and outcome 50) (Figure 1). We grouped variables that were judged to share common mechanisms in their association with sleep and/or LBP, resulting in six domains overall: age, psychological/occupational factors, smoking habits, body mass index, general health, and clinical LBP characteristics. We 28 regarded a study to have controlled for a domain when at least one variable from the domain was considered. We rated a study as having ‘adequate control’ when the study adjusted or controlled for all six domains. We rated a study as having ‘minimal control’ when at least age AND psychological/occupational factors were controlled. These two domains were chosen because there is more robust evidence to support their relationship with sleep and LBP outcomes 40–44. We rated a study as ‘inadequate control’ when age, psychological/occupational factors were not controlled. Studies with inadequate control were rated as high risk of bias in the study confounding domain, those with minimal control were rated as moderate risk of bias and those with adequate control were rated as low risk of bias. Figure 1. Framework for the potential confounders of the association between sleep and low back pain outcomes. Predefined potential confounders were age, psychological/occupational factors (e.g., anxiety, depression, catastrophizing, job satisfaction, work status), smoking habits, body mass index, general health (e.g., physical activity level, comorbidities), and clinical low back pain characteristics (e.g., baseline pain intensity, baseline disability, low back pain duration). Figure created by the authors. Data analyses We used Cohen’s Kappa coefficient to report inter-rater agreement during the study selection process. We used descriptive analysis to summarize the studies’ characteristics and presented them in a 29 descriptive table. LBP duration was categorized as acute LBP (ALBP) (symptoms for less than 12 weeks), chronic LBP (CLBP) (symptoms for 12 weeks or more), and mixed. We categorized studies according to age as younger adults (18-59 years old), older adults (≥60 years old), and mixed. When age range was not available, we considered standard deviations and interquartile intervals to judge which category the study would fall into. Follow-up duration was categorized as short-term (closest to 3 months), moderate-term (closest to 6 months), long-term (closest to 12 months), and very long-term (more than 16 months). We used a ‘synthesis without meta-analysis’ vote-counting approach to summarize the number of studies that found positive, null, or negative associations for each outcome of interest. Among sleep measures, sometimes higher scores/values mean worse sleep (e.g., PSQI score) and sometimes higher scores/values mean better sleep (e.g., total sleep time). Therefore, to report directions of effect, we standardized as a positive association when worse sleep was associated with worse LBP outcomes. When data were sufficiently homogeneous regarding follow-up duration, exposure domain (i.e. ‘baseline sleep’ or ‘changes in sleep’), and adjustment for potential confounders (i.e., unadjusted or adjusted effect sizes), we synthesized outcomes as ‘overall LBP improvement’ outcome (‘improvement’ versus ‘non-improvement’) and all sleep domains as ‘good sleep’ versus ‘poor sleep’ (studies evaluating baseline sleep) or ‘improvement in sleep’ versus ‘non-improvement in sleep’ (studies evaluating changes in sleep) exposures. For studies that measured both pain intensity and disability, we prioritized pain intensity data as previous evidence indicates that no pain is a better measure of feeling recovered than no disability in individuals with LBP 51. When multiple sleep measures were available in a study, we prioritized them according to the order described in the ‘Prognostic factors’ section. We ran random-effects generic inverse variance meta-analysis models in Review Manager 5.4.1 software to investigate the association between sleep (baseline or changes) and overall LBP improvement. We ran separate meta-analyses for unadjusted and adjusted effect sizes, and for short to moderate term (3-6 months) and long to very long term (≥12 months) follow-up periods. When a study reported more than one adjusted effect, we chose the model with the highest number of covariates to pool in our meta-analysis. We calculated unadjusted ORs from studies 30 presenting the raw data and not reporting unadjusted effect sizes. We converted regression coefficients, correlation coefficients, and odds ratios (ORs) into natural log ORs, and synthesized the natural log ORs and standard errors (SEs) to generate pooled ORs and 95% CI 52,53. When the risk ratio (RR) was provided, we pooled them separately. We interpreted effect sizes as small (OR<1.5; RR<1.2), moderate (OR=1.5-2.0; RR=1.2-1.8), or large (OR>2.0; RR>1.8) 54,55. When effect sizes were reported separately for relevant subgroups within a study, e.g., women and men, we used a weighted estimate to pool the effect sizes to generate an estimate for the entire sample. We used the I² value to verify the proportion of the observed dispersion in effect size due to between- studies heterogeneity. We interpreted an I² value above 50% as a significant proportion of dispersion explained by heterogeneity 56,57. Sensitivity analyses We ran three sensitivity analyses to explore the robustness of our results: 1. Limiting to studies with chronic/mixed LBP durations, 2. Limiting to studies with follow-up durations of <24 months (considered to have reasonable biological plausibility for associations between baseline sleep and LBP outcomes), and 3. Limiting to studies using validated sleep measures. Due to insufficient available data, we were unable to perform other previously planned subgroup (e.g., ALBP vs CLBP; younger vs older adults; self-reported vs objective sleep measures) and sensitivity analyses (e.g., influence of studies with high risk of bias and inadequate or minimal control). Assessment of the quality of the evidence We evaluated the quality of the evidence using an adapted version of the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach for prognostic studies 58. We judged the quality of evidence as high, moderate, low, or very low, downgraded based on judgment of the following domains: phase of investigation (most evidence from exploratory studies), study limitations (most evidence from studies with high risk of bias), inconsistency (large I² values, high variability in the direction of association, or minimal overlap of confidence intervals), 31 indirectness (when the sample, prognostic factor and/or outcome of the studies did not accurately reflect the review question), imprecision (insufficient sample size or very wide confidence intervals), publication bias (assuming that prognostic research is likely to be affected by publication bias unless there is strong evidence to the contrary 58). Single studies (not meeting the imprecision criteria) were considered inconsistent and imprecise (i.e., sparse data), providing ‘low-quality evidence’, and were further downgraded to ‘very-low-quality’ if rated as high risk of bias. Evidence of moderate-large effect size (pooled effects of the meta-analysis is moderate or large, or moderate or large similar effects reported by most studies), or exposure-response gradient were factors that could upgrade the quality of evidence. RESULTS Search results Our database search yielded 1,639 records after removing duplicates; we excluded 1,516 at the title/abstract stage. We assessed 123 records in full text and 15 records met our inclusion criteria representing 13 unique studies. The reasons for exclusion during full-text screening are provided in the flowchart (Figure 2). One additional study was identified in our supplemental search and met our inclusion criteria. A total of 14 unique studies from 16 records were included in this review 18,19,67–72,59– 66. Cohen’s Kappa was 0.47 for title/abstract screening and 0.28 for full-text screening. 32 Figure 2. Flowchart of the review selection process. Characteristics of the included studies Table 1 describes the characteristics of each included study. All studies were published between 2014 and 2022, conducted between 1995 and 2018 (unclear in 3 studies) in Sweden (3 studies) 63,66,68, Australia (2 studies) 18,69,71,72, Germany (2 studies) 59, Brazil (1 study) 60,61, Finland (1 study) 67, Spain (1 study) 19, Iran (1 study) 65, Japan (1 study) 64, Norway (1 study) 62, and USA (1 study) 70. Ten studies were prospective cohort studies 19,60,69,61–68 and four studies were secondary analyses of randomized controlled trials 18,59,70. Nine studies were confirmatory studies 18,19,59–62,64,65,67 and five studies were exploratory studies 63,66,68–70. Population: Baseline sample sizes ranged from 129 to 7,164 and totaled 19,170 adults with LBP. Participants were recruited from the general population 59–63,69,70, primary care settings 18,19,66, tertiary care settings 19,65, occupational settings 67,68, and one study recruited survivors from an earthquake 64. 33 The sample was composed of participants with ALBP in two studies 18,69, CLBP in six studies 60– 63,65,66,70, and mixed LBP durations in three studies 19,59 (unclear in 3 studies 64,67,68). The mean or median age ranged from 30 to 71 years old (unclear in 3 studies) and the overall median was 46.0 years old (IQR=41.1, 49.0). Nine studies included only younger adults 18,59,62,65–69, one study included only older adults 60,61, and two studies mixed younger and older adults 19,70 (unclear in 2 studies 63,64). The proportion of female participants ranged from 0 to 100% (unclear in 1 study), and the overall median was 61.0% (IQR=49.7, 71.6). Prognostic factors: The sleep domains of interest were sleep quality (11 studies) 18,19,70,59–61,65–69, sleep quantity (2 studies) 66,69, daytime sleepiness (2 studies) 62,63, and general insomnia symptoms (2 studies) 62,64. Outcomes: Five studies evaluated pain intensity as an outcome 18,19,59,61, four studies evaluated disability 19,60,66,70, and seven studies evaluated recovery 62–65,67–69. Follow-up duration ranged from 3 to 156 months and the median was 6 months (IQR=3, 24). Effect sizes for each study are described in Supplementary Table 1, Appendix A. 34 Table 1. Characteristics of the included studies. Study ID Study design Phase of investigation Setting LBP duration Mean (SD) or median [IQR] age Sleep domain (measure) Outcome (measure) Follow- up (months) Sample size at follow-up Alsaadi 2014 Secondary analysis of an RCT Confirmatory Primary care Acute 44.2 (15.7) Sleep quality (PSQI subscale) Pain intensity (NRS) 3 1,246 Lovgren 2014 Prospective cohort study Exploratory Occupational Unclear Unclear Sleep quality (single question) Recovery (single question) 14 Unclear Lusa 2015 Prospective cohort study Confirmatory Occupational Unclear 37 (6) Sleep quality (single question) Recovery (Nordic Musculoskeletal Questionnaire) 156 38 Nordeman 2017 Prospective cohort study Exploratory Primary care Chronic 45 (10) Sleep quality (single question) Sleep quantity (single question) Disability (RMQ) 24 115 Kovacs 2018 Prospective cohort study Confirmatory Primary care Tertiary care Mixed 48 [28, 64]* 46 [26, 64] Sleep quality (PSQI) Pain intensity (VAS) Disability (RMQ) 3 250† 224 35 53 [30, 64] 49 [29, 64] 220 194 Pakpour 2018 Prospective cohort study Confirmatory Tertiary care Chronic 41.1 (12.2) Sleep quality (PSQI) Recovery (Global Rating of Change Scale and VAS) 6 682 Yabe 2018 Prospective cohort study Confirmatory Survivors from an earthquake Unclear Unclear General insomnia symptoms (Athens Insomnia Scale) Recovery (unclear) 12 535 Halonen 2019 Prospective cohort study Exploratory General population Chronic Unclear Daytime sleepiness (Karolinska Sleep Questionnaire) Recovery (single question) 24 5,740 Klyne 2019# Klyne 2018 Klyne 2020 Prospective cohort study Exploratory General population Acute 30 (8) Sleep quality (PSQI) Sleep quantity (PSQI subscale) Recovery (NRS and RMQ) 6 99 Priebe 2020a Secondary analysis of an RCT Confirmatory General population Mixed 34.0 (10.9) Sleep quality (NRS) Pain intensity (NRS) 3 180 36 Priebe 2020b Secondary analysis of an RCT Confirmatory General population Mixed 47.0 (13.1) Sleep quality (NRS) Pain intensity (NRS) 3 153 Skarpsno 2020 Prospective cohort study Confirmatory General population Chronic 49.1 (11) Insomnia symptoms (single question) Daytime sleepiness (single question) Recovery (Nordic Musculoskeletal Questionnaire) 132 6,200 Roseen 2021 Secondary analysis of an RCT Exploratory General population Chronic 46.1 (10.7) Sleep quality (PSQI) Disability (RMQ) 3 299 Morelhão 2022§ Oliveira 2022 Prospective cohort study Confirmatory General population Chronic 71 (7.5) Sleep quality (PSQI) Pain intensity (NRS) Disability (RMQ) 6 215 IQR=interquartile range; LBP=low back pain, PSQI=Pittsburgh Sleep Quality Index; NRS=Numerical Rating Scale; RCT=randomized controlled trial; RMQ=Roland Morris Questionnaire; SD=standard deviation; VAS=visual analogue scale. * 48 [28, 64] for the association between baseline sleep and pain intensity, 46 [26, 64] for the association between changes in sleep and pain intensity, 53 [30, 64] for the association between baseline sleep and disability, 49 [29, 64] for the association between changes in sleep and disability. † 250 for the association between baseline sleep and pain intensity, 224 for the association between changes in sleep and pain intensity, 220 for the association between baseline sleep and disability, 194 for the association between changes in sleep and disability # Primary report – linked publications did not provide additional data for analysis § Primary report - linked publications provided additional data for analysis 37 Risk of bias assessment Thirteen studies were rated as having high risk of bias and one as having some concerns (Supplementary Table 2, Appendix A). The domains with the highest frequency of high of bias rating were study attrition (9 studies), study confounding (9 studies), study participation (7 studies), and prognostic factor measurement (7 studies). High risk of bias from study attrition was mainly due to low response rates (<75%) and/or poor descriptions of baseline characteristics of those who were lost to follow-up. High risk of bias from study confounding was mainly due to the lack of adjustment/control for potential confounders (Figure 1). High risk of bias from study participation was mainly due to poor reporting of participants characteristics such as LBP duration, baseline LBP severity, and lack of definition of what was considered as non-specific LBP. The high risk of bias from prognostic factor measurement was mainly due to the use of non-validated sleep measures. Sleep as a prognostic factor for pain intensity outcomes Three studies investigated the association between baseline sleep and future pain intensity 18,19,61, including 1,711 participants with follow-up data. One study provided both unadjusted and adjusted effect sizes 18, one provided only unadjusted effect sizes 19 and another one provided only adjusted effect sizes 61. Two studies found positive associations between baseline sleep quality and pain intensity. One at a 3-month follow-up in younger adults with ALBP 18 and another one at a 6-month follow-up in older adults with CLBP 61. One study found no association between sleep quality and pain intensity at a 3-month follow-up in a mixed sample of younger and older adults and mixed LBP durations 19. We found very low-quality evidence (Supplementary Table 3, Appendix A) of a positive association between baseline sleep and future pain intensity (Figure 3a). Sleep as a prognostic factor for disability outcomes Four studies investigated the association between baseline sleep and future disability 19,60,66,70, totaling a sample of 849 participants with follow-up data. Three studies provided only unadjusted effect sizes 19,66,70 and one study provided only adjusted effect sizes 60. For one study 70, we could extract only the 38 raw data (i.e., the number of participants with good and poor sleep at baseline in the improved and not improved groups); thus, we calculated unadjusted ORs to report the results. Two studies reported positive associations between baseline sleep quality and disability a 3-month follow-up in a mixed sample of younger and older adults with CLBP 70, and at a 6-month follow-up in older adults with CLBP 73. One study found no association between baseline sleep quality and disability at a 3-month follow-up in a mixed sample of younger and older adults, mixed LBP durations 19, and one study found no association between baseline sleep quality and sleep quantity with percentage of improvement in disability at a 24-month follow-up in younger adults with CLBP 66. We found very low-quality evidence (Supplementary Table 3, Appendix A) of no association between baseline sleep and future disability (Figure 3b). Sleep as a prognostic factor for recovery outcomes Six studies evaluated the association between baseline sleep and recovery of LBP 62,63,65,67–69. We were unable to extract the final sample size from one study and it was not used in our data synthesis 68; thus, the remaining 5 studies 62,63,65,67,69 totaled 13,294 participants with follow-up data. Two studies provided only unadjusted effects 67,69, one study provided only adjusted effects 63, and two studies provided both unadjusted and adjusted effects 62,65. One study 62 did not report unadjusted effects but we calculated unadjusted ORs and RRs from the raw data presented in the article. Similarly, we calculated unadjusted ORs and RRs from the raw data reported in another study 67 considering only the recovery categories that we could assume had LBP at baseline (i.e., ‘recovering pain’ and ‘chronic pain’ categories). Three studies found positive associations between sleep and recovery. One study 65 found a positive association between baseline sleep quality and recovery in younger adults with CLBP at a 6-month follow-up. Another study 63 found a positive association between baseline daytime sleepiness and recovery at a 24-month follow-up in individuals (unclear whether younger or older adults) with CLBP. In one study 62, having ‘1’, ‘2’, or ‘3’ insomnia symptoms were positively associated with recovery at a 132-month follow-up in younger adults with CLBP. In the same study, having daytime sleepiness symptoms ‘sometimes’ and ‘often/always’ were also positively associated with recovery. The authors further investigated whether having pain in other body regions was an 39 effect modifier of the association between baseline sleepiness and LBP recovery and no effect modification was found. There was no association between baseline sleep and recovery in two studies. In one study 67, having ‘mild’ or ‘severe’ poor sleep quality was not associated with recovery at a 156- month follow-up in younger adults with LBP (unclear duration). In another study 69, there was no difference in mean sleep quality and mean sleep quantity between recovery categories at a 6-month follow-up in younger adults with ALBP. We found very low-quality evidence (Supplementary Table 3, Appendix A) of a positive association between baseline sleep and recovery (Figure 3c). Figure 3. Graphs illustrating our vote-counting approach with the number of studies, their respective sample sizes, and reported associations (positive, no association, or negative) for baseline sleep and outcomes: a. future pain intensity, b. disability, and c. recovery. Each bar represents a sleep domain evaluated by an individual study; the bar height represents the study sample size. Bars in black 40 represent a ‘positive association’ and gray bars represent ‘no association’. No study found a negative association. *=Studies that evaluated two sleep domains are represented twice. Sleep as a prognostic factor for overall LBP improvement Nine studies provided usable data on the association between baseline sleep and overall LBP improvement to be included in our data synthesis 18,19,61–63,65,67,70. Four studies (2,477 participants) reported unadjusted effect sizes for short-moderate term follow-up (Figure 4a) 18,19,65,70, three studies (2,143 participants) reported adjusted effect sizes for short-moderate term follow-up (Figure 4b)18,61,65, three studies (6,353 participants) reported unadjusted effect sizes for long-very long term follow-up (Figure 5a) 62,66,67, and two studies (11,940 participants) reported adjusted effect sizes for long-very long follow-up (Figure 5b) 62,63. We found very low-quality evidence (Supplementary Table 3, Appendix A) of no association between sleep and overall LBP improvement in the short-moderate term. We found very low-quality evidence (Supplementary Table 3, Appendix A) that poor sleep was moderately associated with non-improvement in LBP in the long-very long term in the pooled unadjusted effects; however, no association was found in the pooled adjusted effects. Figure 4. Forest plot of the unadjusted (4a) and adjusted (4b) associations between baseline sleep and chance of non-improvement in short-moderate term (3 to 6 months of follow-up). 41 Figure 5. Forest plot of the unadjusted (5a) and adjusted (5b) associations between baseline sleep and chance (5a)/ risk (5b) of non-improvement in long-very long term (≥12 months of follow-up). Association between changes in sleep and changes in pain intensity Three studies presented data on the association between changes in sleep and changes in pain intensity, totaling a sample of 557 participants with follow-up data 19,59. All studies mixed participants with ALBP and CLBP. All studies provided unadjusted effect sizes and found positive associations between changes in sleep quality and changes in pain intensity at a 3-month follow-up in younger adults 59 and in a mixed sample of younger and older adults with mixed LBP durations 19. Therefore, there was low-quality evidence (Supplementary Table 3, Appendix A) of a positive association between changes in sleep and changes in pain intensity. Association between changes in sleep and changes in disability One study evaluated the association between changes in sleep quality and changes in disability, totaling a sample size of 194 participants with follow-up data 19. The study found a positive association between improvement in sleep and improvement in disability at a 3-month follow-up in a mixed sample of younger and older adults, mixed LBP durations 19. Therefore, there was very low- quality evidence (Supplementary Table 3, Appendix A) of a positive association between changes in sleep and changes in disability. Association between changes in sleep and recovery Two studies (1,217 participants with follow-up data) evaluated the association between changes in sleep and recovery, and both found positive associations 64,65. In one study 64, continuation of insomnia symptoms was associated with non-recovery at a 12-month follow-up. Sample age and LBP duration were unclear in this study. Another study 65 found associations between the ‘development’ of poor 42 sleep quality and ‘persistent’ poor sleep quality with non-recovery at a 6-month follow-up in younger adults with CLBP. Therefore, there was low-quality evidence (Supplementary Table 3, Appendix A) of a positive association between changes in sleep and recovery. Association between changes in sleep and overall LBP improvement Four studies provided usable data on the association between changes in sleep and overall LBP improvement to be included in our quantitative synthesis 19,59,65. All studies provided unadjusted effect sizes for short-moderate term follow-up (1,239 participants). We found low-quality evidence (Supplementary Table 3, Appendix A) of a large association between non-improvement in sleep and non-improvement in LBP outcomes in the short-moderate term (Figure 6). Figure 6. Forest plot of the unadjusted association between changes in sleep and chance of non- improvement in low back pain outcomes in short-moderate term (3 to 6 months of follow-up). Sensitivity analyses When limiting to studies with chronic/mixed LBP durations, there was a shift from a positive association to no association between baseline sleep and future pain intensity. Limiting to studies with <24 months of follow-up resulted in changes from null to a positive association between baseline sleep and future disability, and from a positive to null association between baseline sleep and recovery. All studies included in the meta-analyses for the long-very long term had follow-ups of ≥24 months. Limiting to studies that used validated sleep measures resulted in changes from a positive to null association between baseline sleep and future pain intensity, and from null to a positive association between baseline sleep and future disability. There was a shift from low-quality to very-low quality of evidence for the association between changes in sleep and changes in pain intensity when limiting to studies that used validated sleep measures. Interpretation of other results was not changed by 43 sensitivity analyses. DISCUSSION Summary of findings We found positive associations between baseline sleep with future pain intensity, recovery, and overall LBP improvement in the long-very long term; and no association between baseline sleep with disability and overall LBP improvement in the short-moderate term. We found positive associations between changes in sleep with changes in pain intensity, disability, recovery, and overall LBP improvement in the short-moderate term. All findings were supported by low or very low-quality of evidence, which means that future studies are likely to change the estimates. In addition, there was high clinical heterogeneity among the studies and a significant proportion of dispersion of effect sizes was explained by heterogeneity (I²>50%). Therefore, the interpretation of our findings must be done with caution. Comparison with the literature and implications for clinical practice Our findings are in line with expert recommendations that clinicians should assess sleep in patients presenting for LBP management 17. Worse baseline sleep seems to be associated with worse LBP outcomes (except for disability). This finding contradicts a previous review that found no association between baseline sleep quality and future CLBP outcomes 74. This divergence can be explained by the broader scope covered by our review and the inclusion of more studies. This previous review only included studies that evaluated sleep at baseline and follow-up, which may limit the generalizability of their conclusions regarding ‘baseline sleep’. Furthermore, we found consistent and large associations between non-improvement in sleep and non-improvement in LBP outcomes. This corroborates Chang et al. (2022), who found relationships between improvement in sleep quality and improvement in CLBP outcomes 74. Therefore, we also recommend clinicians consider managing sleep problems (or referral to a specialist if needed) in conjunction with LBP management. Again, interpretation must be done with caution, considering the low and very-low certainty of the evidence, and that findings came 44 substantially from inadequately adjusted effects in which confounding may explain some associations found. Limitations of the included studies and recommendations for future studies No study met our pre-defined criteria for adequate control for potential confounders. We encourage future prognostic studies to pre-define all potential confounders when designing their studies. Furthermore, non-validated sleep measures were used in seven studies, and some of our findings were impacted when we limited to studies using valid measures. Non-validated measures may not capture sleep adequately and may introduce measurement bias. Future studies should use structured and valid measures. We identified that sleep quality has been the most investigated sleep domain in the field. Sleep quality is a complex construct that integrates factors such as sleep quantity, sleep fragmentation, feeling restored, time spent in deep sleep phases 21,22,75. Most of the evidence investigating the mechanisms that explain how sleep seems to influence pain processing comes from sleep deprivation studies 14, however, sleep quantity has been understudied as a prognostic factor in LBP. We identified only two exploratory studies that investigated sleep quantity, and both used non-validated sleep measures. Future confirmatory prognostic studies are needed to investigate the role of sleep quantity as a prognostic factor in LBP. We acknowledge that the gold standard for measuring sleep quantity (i.e., polysomnography and actigraphy 24) may not be feasible to be implemented in large studies or clinical practice as they have high costs and require specialized professionals. If not feasible, prospective sleep diaries recording at least 7 days are preferred self-reported measures of sleep quantity 76. We found only one confirmatory study with only ALBP. This study found the strongest association observed between baseline sleep and future LBP outcomes. Limiting to studies with chronic/mixed LBP durations changed the interpretation of some of our results. This may suggest a stronger relationship between sleep and LBP outcomes in ALBP and may indicate a greater need for sleep assessment in this population. However, this study was rated as high risk of bias and minimally 45 controlled for confounders. Future high-quality studies with ALBP are needed to try to replicate these findings. Limitations and strengths of our review We included and pooled studies evaluating a variety of sleep domains which contributed the observed heterogeneity. We included all these sleep domains to allow broad assessment in this growing area of research, and to make recommendations for future studies. Furthermore, we included studies using non-validated tools to measure sleep. The inconsistent use of sleep measures is a known issue in the field 8, thus, we knew in advance that only accepting studies using valid measures would severely restrict the amount of usable data for synthesis. Additionally, we acknowledge the high potential for publication bias and selective outcome reporting bias in the field, as prospective registration is not mandatory for the publication of observational studies. This may have led to an overestimation of strength of the associations found. Another limitation was the mix of ALBP and CLBP in our analyses. We had planned a subgroup analysis separating acute from CLBP; however, the small number of studies with ALBP prevented this. It is also noteworthy that the pooled adjusted effect sizes came from studies that adjusted for different covariates; thus, interpretation of the results from these estimates must be done with caution. Strengths of our study include our comprehensive database and supplemental search approaches, all recommended for reviews of prognostic factor studies 77; applying no restriction on language of publication to our search; conducting a GRADE assessment for each association of interest; and the mix of meta-analyses with synthesis without meta-analysis methods. Our results suggest that sleep may be associated with future LBP outcomes (except disability) and non-improvement in sleep may be associated with non-improvement in LBP. However, these findings were supported by low to very low-quality of evidence and better-conducted studies are needed to strengthen our certainty about the evidence. 46 ACKNOWLEDGMENTS: the authors would like to thank the Pro-Reitoria de Pesquisa (PRPq) at Universidade Federal de Minas Gerais (UFMG), Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG), Conselho Nacional de Desenvolvimento Científico e Tecnológio (CNPq), Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Centro de Estudos em Psicobiologia e Exercício (CEPE), Fundação de Apoio ao Ensino, Pesquisa e Extensão (FEPE/UFMG), and Global Affairs Canada. The authors would also like to thank Mrs. Rachel Ogilvie from Dalhousie University for the support on the manuscript writing. FUNDING SOURCE: none CONFLICT OF INTEREST: none. 47 REFERENCES [1] Alsaadi SM, McAuley JH, Hush JM, Lo S, Lin C-WC, Williams CM, Maher CG. Poor sleep quality is strongly associated with subsequent pain intensity in patients with acute low back pain. Arthritis Rheumatol 2014;66:1388–1394. [2] Altman DG, Bland JM. How to obtain the confidence interval from a P value. BMJ 2011;343:d2090. [3] Artus M, Campbell P, Mallen CD, Dunn KM, Van Der Windt DAW. Generic prognostic factors for musculoskeletal pain in primary care: A systematic review. BMJ Open 2017;7:1–10. [4] Azevedo E, Manzano GM, Silva A, Martins R, Andersen ML, Tufik S. The effects of total and REM sleep deprivation on laser-evoked potential threshold and pain perception. Pain 2011;152:2052–2058. doi:10.1016/j.pain.2011.04.032. [5] Bastien CH, Vallières A, Morin CM. Validation of the Insomnia Severity Index as an outcome measure for insomnia research. Sleep Med 2001;2:297–307. [6] Borba DDA, Reis RS, Lima PH, Facundo LA, Narciso FV, Silva A, de Mello MT. How many days are needed for a reliable assessment by the Sleep Diary? Sleep Sci 2020;13:49–53. [7] Borenstein M, Hedges L V, Higgins JP, Rothstein HR. Converting Among Effect Sizes. Introduction to Meta-Analysis.2009. pp. 45–49. [8] Borenstein M, Hedges L V, Higgins JP, Rothstein HR. Introduction to Meta-Analysis. 1st ed. Chichester: John Wiley & Sons, 2009. [9] Boulos L, Ogilvie R, Hayden JA. Search methods for prognostic factor systematic 48 reviews: a methodologic investigation. J Med Libr Assoc 2021;109:23–32. [10] Buysse DJ, Reynolds CF, Monk TH, Berman SR, Kupfer DJ. The Pittsburgh Sleep Quality Index: A New Instrument Psychiatric Practice and Research. Psychiatry Res 1989;28:193–211. [11] Cappuccio FP, Taggart FM, Kandala NB, Currie A, Peile E, Stranges S, Miller MA. Meta-analysis of short sleep duration and obesity in children and adults. Sleep 2008;31:619–626. [12] Chang JR, Wang X, Lin G, Samartzis D, Pinto SM, Wong AYL. Are Changes in Sleep Quality / Quantity or Baseline Sleep Parameters Related to Changes in Clinical Outcomes in Patients With Nonspecific Chronic Low Back Pain? A Systematic Review. Clin J Pain 2022;38:292–307. [13] Coombs DM, Machado GC, Richards B, Wilson R, Chan J, Storey H, Maher CG. Healthcare costs due to low back pain in the emergency department and inpatient setting in Sydney, Australia. Lancet Reg Heal - West Pacific 2021;7:100089. [14] Costa LOP, Maher CG, Latimer J, Ferreira PH, Ferreira ML, Pozzi GC, Freitas LMA. Clinimetric testing of three self-report outcome measures for low back pain patients in Brazil: Which one is the best? Spine (Phila Pa 1976) 2008;33:2459–2463. [15] Dieleman JL, Cao J, Chapin A, Chen C, Li Z, Liu A, Horst C, Kaldjian A, Matyasz T, Scott KW, Bui AL, Campbell M, Duber HC, Dunn AC, Flaxman AD, Fitzmaurice C, Naghavi M, Sadat N, Shieh P, Squires E, Yeung K, Murray CJL. US Health Care Spending by Payer and Health Condition, 1996-2016. JAMA - J Am Med Assoc 2020;323:863–884. 49 [16] Fairbank J, Couper J, Davies J, O’Brien J. The Oswestry low back pain disability questionnaire. Physiotherapy 1980;66:271–273. [17] Ferreira ML, Luca K de, Haile LM, Steinmetz JD, Culbreth GT, Cross M, Kopec J, Ferreira PH, Blyth FM, Buchbinder R, Hartvigsen J, Wu A-M, Safiri S, Woolf A, Collins GS, Ong KL, Fukutaki KG. Global, regional, and national burden of low back pain, 1990 – 2020, its attributable risk factors, and projections to 2050: a systematic analysis of the Global Burden of Disease Study 2021. Lancet Rheumatol 2023;5:e316– e329. [18] Finan PH, Goodin BR, Smith MT. The Association of Sleep and Pain: An Update and a Path Forward. J Pain 2016;14:1539–1552. [19] Halonen JI, Shiri R, Magnusson Hanson LL, Lallukka T. Risk and Prognostic Factors of Low Back Pain: Repeated Population-based Cohort Study in Sweden. Spine (Phila Pa 1976) 2019;44:1248–1255. [20] Hayden JA, Wilson M, Riley R, Iles R, Pincus T, Ogilvie R. Individual recovery expectations and prognosis of outcomes in non- specific low back pain: prognostic factor review. Cochrane Database 2019;11:CD011284. [21] Hayden JA, Windt DA Van Der, Cartwright JL, Co P. Assessing Bias in Studies of Prognostic Factors. Ann Behav Med 2006;158:280–286. [22] Higgins JP, Thomas J, Chandler J, M C, T L, MJ P, VA W (editors). Cochrane Handbook for Systematic Reviews of Interventions version 6.3 (updated February 2022). Cochrane, 2022 2022. [23] Huguet A, Hayden JA, Stinson J, Mcgrath PJ, Chambers CT, Tougas ME, Wozney L. 50 Judging the quality of evidence in reviews of prognostic factor research: adapting the GRADE framework. Syst Rev 2013;2:71. [24] Johns MW. A New Method for Measuring Daytime Sleepiness: The Epworth Sleepiness Scale. Sleep 1991;14:540–545. [25] Kaida K, Takahashi M, Akerstedt T, Nakata A, Otsuka Y, Haratani T, Fukasawa K. Validation of the Karolinska sleepiness scale against performance and EEG variables. Clin Neurophysiol 2006;117:1574–1581. [26] Kamper SJ, Maher CG, Herbert RD, Hancock MJ, Hush JM, Smeets RJ. How little pain and disability do patients with low back pain have to experience to feel that they have recovered? Eur Spine J 2010;19:1495–1501. [27] Kamper SJ, Maher CG, Mackay G. Global Rating of Change Scales: A Review of Strenghts and Weaknesses and Considerations for Design. J Man Manip Ther 2009;17:163–170. [28] Kline C. Sleep Quality. Encyclopedia of Behavioral Medicine.2013. p. 1797. [29] Klyne DM, Barbe MF, Hoorn W Van Den, Hodges PW. ISSLS PRIZE IN CLINICAL SCIENCE 2018 : longitudinal analysis of inflammatory, psychological, and sleep ‑ related factors following an acute low back pain episode — the good, the bad, and the ugly. Eur Spine J 2018;27:763–777. [30] Klyne DM, Hodges PW. Circulating Adipokines in Predicting the Transition from Acute to Persistent Low Back Pain. Pain Med 2020;0:1–11. [31] Klyne DM, Moseley GL, Sterling M, Barbe MF, Hodges PW. Are Signs of Central Sensitization in Acute Low Back Pain a Precursor to Poor Outcome? J Pain 51 2019;20:994–1009. [32] Kovacs FM, Seco J, Royuela A, Betegon JN, Sanchez-Herraez S, Meli M, Martinez Rodriguez ME, Nunez M, Alvarez-Galovich L, Moya J, Sanchez C, Luna S, Borrego P, Moix J, Rodriguez-Perez V, Torres-Unda J, Burgos-Alonso N, Gago-Fernandez I, Gonzalez-Rubio Y, Abraira V. The association between sleep quality, low back pain and disability: A prospective study in routine practice. Eur J Pain 2018;22:114–126. [33] Krishnan V, Dixon-Williams S, Thornton JD. Where There Is Smoke…There Is Sleep Apnea: Exploring the Relationship Between Smoking and Sleep Apnea. Chest 2014;146:1673–1680. [34] Kuorinka I, Jonsson B, Kilbom A, Vinterberg H, Biering-Sørensen F, Andersson G, Jørgensen K. Standardised Nordic questionnaires for the analysis of musculoskeletal symptoms. Appl Ergon 1987;18:233–237. [35] Larson RA, Carter JR. Total sleep deprivation and pain perception during cold noxious stimuli in humans. Scand J Pain 2016;13:12–16. [36] Lautenbacher S, Peters JH, Heesen M, Scheel J, Kunz M. Age changes in pain perception: A systematic-review and meta-analysis of age effects on pain and tolerance thresholds. Neurosci Biobehav Rev 2017;75:104–113. [37] Lavigne GJ, Nashed A, Manzini C, Carra MC. Does sleep differ among patients with common musculoskeletal pain disorders? Curr Rheumatol Rep 2011;13:535–542. [38] Lehrer HM, Yao Z, Krafty RT, Evans MA, Buysse DJ, Kravitz HM, Matthews KA, Gold EB, Harlow SD, Samuelsson LB, Hall MH. Comparing polysomnography, actigraphy, and sleep diary in the home environment: The Study of Women’s Health 52 Across the Nation (SWAN) Sleep Study. SLEEP Adv 2022;3:1–12. [39] Lovgren M, Gustavsson P, Melin B, Rudman A. Neck/shoulder and back pain in new graduate nurses: A growth mixture modeling analysis. Int J Nurs Stud 2014;51:625– 639. [40] Lusa S, Miranda H, Luukkonen R, Punakallio A. Sleep disturbances predict long-term changes in low back pain among Finnish firefighters: 13-year follow-up study. Int Arch Occup Environ Health 2015;88:369–379. [41] Mahdavi SB, Riahi R, Vahdatpour B, Kelishadi R. Association between sedentary behavior and low back pain; A systematic review and meta-analysis. Heal Promot Perspect 2021;11:393–410. [42] Matthews KA, Patel SR, Pantesco EJ, Buysse DJ, Kamarck TW, Lee L, Hall MH. Similarities and differences in estimates of sleep duration by polysomnography, actigraphy, diary, and self-reported habitual sleep in a community sample. Sleep Heal J Natl Sleep Found 2017;4:96–103. [43] Melzack R. The McGill Pain Questionnaire: Major properties and scoring methods. Pain 1975;1:277–299. [44] Morelhão PK, Gobbi C, Christofaro DGD, Damato TM, Grande GD, Frange C, Andersen ML, Tufik S, Franco MR, Pinto RZ. Bidirectional Association Between Sleep Quality and Low Back Pain in Older Adults: A Longitudinal Observational Study. Arch Phys Med Rehabil 2022;103:1558–1564. [45] Morelhão PK, Kim LJ, Pinto RZ, Tufik S, Andersen ML. Should Physical Therapists Assess Sleep Quality in Patients Seeking Care for Low Back Pain. Phys Ther 53 2019;99:961–963. [46] Nelson KL, Davis JE, Corbett CF. Sleep quality: An evolutionary concept analysis. Nursin Forum 2022;57:144–151. [47] Nieminen KL, Pyysalo ML, Kankaanpää JM. Prognostic factors for pain chronicity in low back pain: a systematic review. Pain 2021;6:e919. [48] Nordeman L, Thorselius L, Gunnarsson R, Mannerkorpi K. Predictors for future activity limitation in women with chronic low back pain consulting primary care: a 2- year prospective longitudinal cohort study. BMJ Open 2017;7:e013974. [49] Ohayon MM, Carskadon MA, Guilleminault C, Vitiello M V. Meta-Analysis of Quantitative Sleep Parameters From Childhood to Old Age in Healthy Individuals: Developing Normative Sleep Values Across the Human Lifespan. Sleep 2004;27:1255–1273. [50] Oliveira SD, Pinto RZ, Gobbi C, Fernandes GL, Dokkedal-Silva V, Lemes IR, Andersen ML, Tufik S, Lorenconi RMR, Morelhao PK. Sleep Quality Predicts Functional Disability in Older Adults with Low Back Pain: A Longitudinal Study. J Appl Gerontol 2022;41:2374–2381. [51] Oliveira SD, Pinto RZ, Gobbi C, Fernandes GL, Dokkedal-Silva V, Lemes ÍR, Andersen ML, Tufik S, Lorenconi RMR, Morelhão PK. Sleep Quality Predicts Functional Disability in Older Adults with Low Back Pain: A Longitudinal Study. J Appl Gerontol 2022;41:2374–2381. [52] Olivier J, May WL, Bell ML. Relative effect sizes for measures of risk. Commun Stat - Theory Methods 2017;46:6774–6781. 54 [53] Page MJ, Mckenzie JE, Bossuyt PM, Boutron I, Hoffmann C, Mulrow CD, Shamseer L, Tetzlaff JM, Akl EA, Brennan SE, Chou R, Glanville J, Grimshaw JM, Hróbjartsson A, Lalu MM, Li T, Loder EW, Mayo-wilson E, Mcdonald S, Mcguinness LA, Stewart LA, Thomas J, Tricco AC, Welch VA, Whiting P, Moher D. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews Systematic reviews and Meta-Analyses. BMJ 2021;372. [54] Pakpour AH, Yaghoubidoust M, Campbell P. Persistent and Developing Sleep Problems: A Prospective Cohort Study on the Relationship to Poor Outcome in Patients Attending a Pain Clinic with Chronic Low Back Pain. Pain Pract 2018;18:79–86. [55] Pinheiro MB, Ferreira ML, Refshauge K, Maher CG, Ordoñana JR, Andrade TB, Tsathas A, Ferreira PH. Symptoms of depression as a prognostic factor for low back pain: A systematic review. Spine J 2016;16:105–116. [56] Priebe JA, Utpadel-Fischler D, Toelle TR. Less Pain, Better Sleep? The Effect of a Multidisciplinary Back Pain App on Sleep Quality in Individuals Suffering from Back Pain - a Secondary Analysis of App User Data. J Pain Res 2020;13:1121–1128. [57] Riley RD, Moons KGM, Snell KIE, Ensor J, Hooft L, Altman DG, Hayden J, Collins GS, Debray TPA. A guide to systematic review and meta-analysis of prognostic factor studies. BMJ 2019;364:k4597. [58] Roland M, Fairbank J. The Roland-Morris Disability Questionnaire and the Oswestry Disability Questionnaire. Spine (Phila Pa 1976) 2000;25:3115–3124. [59] Roseen EJ, Gerlovin H, Felson DT, Delitto A, Sherman KJ, Saper RB. Which Chronic Low Back Pain Patients Respond Favorably to Yoga, Physical Therapy, and a Self-care Book? Responder Analyses from a Randomized Controlled Trial. Pain Med 55 2021;22:165–180. [60] Sateia MJ. International classification of sleep disorders - third edition: highlights and modifications. Chest 2014;146:1387–1394. [61] Scott H, Lack L, Lovato N. A systematic review of the accuracy of sleep wearable devices for estimating sleep onset. Sleep Med Rev 2020;49:101227. [62] Shrivastava D, Jung S, Saadat M, Sirohi R, Crewson K. How to interpret the results of a sleep study. J Community Hosp Intern Med Perspect 2014;4:1–4. [63] Silva A, Mello MT, Serrão PR, Luz RP, Ruiz F, Bittencourt LR, Tufik S, Mattiello SM. Influence of Obstructive Sleep Apnea in the Functional Aspects of Patients With Osteoarthritis. J Clin Sleep Med 2018;14:265–270. [64] Simpson NS, Scott-Sutherland J, Gautam S, Sethna N, Haack M. Chronic exposure to insufficient sleep alters processes of pain habituation and sensitization. Pain 2018;159:33–40. [65] Skarpsno ES, Mork PJ, Nilsen TIL, Nordstoga AL. Influence of sleep problems and co- occurring musculoskeletal pain on long-term prognosis of chronic low back pain: the HUNT Study. J Epidemiol Community Health 2020;74:283–289. [66] Soldatos CR, Dikeos DG, Paparrigopoulos TJ. Athens Insomnia Scale : validation of an instrument based on ICD-10 criteria. J Psychosom Res 2000;48:555–560. [67] Sun Y, Laksono I, Selvanathan J, Saripella A, Nagappa M, Pham C, Englesakis M, Peng P, Morin CM, Chung F. Prevalence of sleep disturbances in patients with chronic non-cancer pain: A systematic review and meta-analysis. Sleep Med Rev 2021;57:101467. 56 [68] Vanderweele TJ. Principles of confounder selection. Eur J Epidemiol 2021;34:211– 219. [69] Whibley D, Alkandari N, Kristensen K, Barnish M, Rzewuska M, Druce KL, Tang NKY. Sleep and Pain: A Systematic Review of Studies of Mediation. Clin J Pain 2019;35:544–558. [70] van der Wurf C, Speklé E, Schaafsma F, Coenen P. Determining the Costs of Low- Back Pain Associated Sick Leave in the Dutch Workforce in the Period 2015 to 2017. J Occup Environ Med 2021;63:e367–e372. [71] Yabe Y, Hagiwara Y, Sekiguchi T, Sugawara Y, Tsuchiya M, Koide M, Itaya N, Yoshida S, Sogi Y, Yano T, Tsuji I, Itoi E. Sleep Disturbance Is Associated with New Onset and Continuation of Lower Back Pain: A Longitudinal Study among Survivors of the Great East Japan Earthquake. Tohoku J Exp Med 2018;246:9–14. [72] Yang Y, Shin JC, Li D, An R. Sedentary Behavior and Sleep Problems: a Systematic Review and Meta-Analysis. Int J Behav Med 2017;24:481–492. [73] Zhang J-M, An J. Cytokines, Inflammation and Pain. Int Anesthesiol Clin 2009;45:27– 37. 57 APPENDIX A The full search strategies were developed with the help of a librarian with expertise in health sciences. MEDLINE (Ovid) 1 exp Sleep/ 2 exp Sleep Wake Disorders/ or exp Sleep-Wake Transition Disorders/ 3 Sleep*.tw,kf. 4 (Hyposomni* or parasomni* or dyssomni*).tw,kf. 5 insomni*.tw,kf. 6 1 or 2 or 3 or 4 or 5 7 exp Back Pain/ 8 Intervertebral Disc Displacement/ 9 exp Sciatic Neuropathy/ 10 exp Spondylosis/ 11 (back ache* or backache* or back disorder* or back pain*).tw,kw,kf. 12 coccydynia.tw,kw,kf. 13 ((disc? or disk?) adj1 (degenerat* or displace* or hernia* or prolapse* or slipped)).tw,kw,kf. 14 dorsalgia.tw,kw,kf. 15 ((lumb* or spin* or vertebr*) adj4 pain).tw,kw,kf. 16 lumbago.tw,kw,kf. 17 (sciatic neuropathy or sciatica or ischialgia).tw,kw,kf. 18 (spondylosis or spondylolysis or spondylolisthesis).tw,kw,kf. 19 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 or 17 or 18 20 exp Cohort Studies/ or incidence.tw,kf. or exp Mortality/ or exp Follow-Up Studies/ or prognos*.tw,kf. or predict*.tw,kf. or course.tw,kf. or cohort*.tw,kf. or exp Survival Analysis/ 21 6 and 19 and 20 58 Embase (www.embase.com) #70. #67 AND #68 AND #69 #69. #59 OR #60 OR #61 OR #62 OR #63 OR #64 OR #65 OR #66 #68. #41 OR #42 OR #43 OR #44 OR #45 OR #46 OR #47 OR #48 OR #49 OR #50 OR #51 OR #52 OR #53 OR #54 OR #55 OR #56 OR #57 OR #58 #67. #36 OR #37 OR #38 OR #39 OR #40 #66. 'survival analysis'/exp #65. course:ti,ab,kw OR cohort*:ti,ab,kw #64. predict*:ti,ab,kw #63. prognos*:ti,ab,kw #62. 'follow up'/exp #61. 'mortality'/exp #60. incidence:ti,ab #59. 'cohort analysis'/exp #58. ((disc OR disk) NEAR/1 (degenerat* OR displace* OR hernia* OR prolapse* OR slipped)):ti,ab,kw #57. spondylolisthesis:ti,ab,kw OR (((lumb* OR spin* OR vertebr*) NEAR/4 pain):ti,ab,kw) #56. spondylolysis:ti,ab,kw #55. spondylosis:ti,ab,kw #54. ischialgia:ti,ab,kw #53. sciatica:ti,ab,kw #52. 'sciatic neuropathy':ti,ab,kw #51. lumbago:ti,ab,kw #50. dorsalgia:ti,ab,kw #49. coccydynia:ti,ab,kw #48. 'back pain*':ti,ab,kw #47. 'back disorder*':ti,ab,kw #46. backache*:ti,ab,kw #45. 'back ache*':ti,ab,kw #44. 'spondylosis'/exp #43. 'sciatic neuropathy'/exp #42. 'intervertebral disk hernia'/exp 59 #41. 'low back pain'/exp #40. insomni*:ti,ab,kw #39. hyposomni*:ti,ab,kw OR parasomni*:ti,ab,kw OR dyssomni*:ti,ab,kw #38. sleep*:ti,ab,kw #37. 'sleep disorder'/exp/mj #36. 'sleep'/exp/mj 60 CINAHL (EBSCO) S20. S15 AND S18 AND S19 S19. S1 OR S2 OR S3 OR S4 OR S5 OR S6 OR S7 S18. S16 OR S17 S17. TI ( sleep* OR parasomni* OR hyposomni* OR insomni* OR dyssomni* ) OR AB ( sleep* OR parasomni* OR hyposomni* OR insomni* OR dyssomni* ) S16. (MH "Sleep Disorders, Intrinsic+") OR (MH "Dyssomnias+") OR (MH "Sleep Disorders+") OR (MH "Sleep Disorders, Circadian Rhythm+") OR (MH "Sleep-Wake Transition Disorders+") OR (MH "Parasomnias+") OR (MH "Sleep+") OR (MH "Sleep Hygiene+") OR (MH "Sleep Stages+") S15. S8 OR S9 OR S10 OR S11 OR S12 OR S13 OR S14 S14. (MH "Prospective Studies+") S13. (MH "Prognosis+") S12. (MH "Survival Analysis+") S11. (TI (predict* OR prognos* OR course OR cohort* or incidence) OR AB (predict* OR prognos* OR course OR cohort* OR incidence)) S10. (TI "follow up stud*" or AB "follow up stud*") S9. (MH "Mortality") S8. (MH "Incidence") S7. TI ("back pain*" OR backache* OR "back ache*") OR AB ("back pain*" OR backache* OR "back ache*") S6. TI (spondylolysis OR spondylolisthesis OR spondylosis OR lumbago OR ischialgia OR dorsalgia OR "sciatic neuropathy" OR sciatica OR coccydynia) OR AB (spondylolysis OR spondylolisthesis OR spondylosis OR lumbago OR ischialgia OR dorsalgia OR "sciatic neuropathy" OR sciatica OR coccydynia) S5. TI ( ((lumb* or spin* or vertebr*) N4 pain) ) OR AB ( ((lumb* or spin* or vertebr*) N4 pain) ) S4. TI ( ((disc or discs or disk or disks) N1 (degenerat* or displace* or hernia* or prolapse* or slipped)) ) OR AB ( ((disc or discs or disk or disks) N1 (degenerat* or displace* or hernia* or prolapse* or slipped)) ) S3. (MH “Spondylosis+”) S2. (MH "Intervertebral Disk Displacement") S1. (MH "Back Pain+") 61 PsycINFO (EBSCO) S15. S7 AND S13 AND S14 S14. S3 OR S4 OR S8 OR S10 S13. S1 OR S2 OR S9 OR S11 OR S12 S12. TI ( ((disc or discs or disk or disks) N1 (degenerat* or displace* or hernia* or prolapse* or slipped)) ) OR AB ( ((disc or discs or disk or disks) N1 (degenerat* or displace* or hernia* or prolapse* or slipped)) ) OR KW ( ((disc or discs or disk or disks) N1 (degenerat* or displace* or hernia* or prolapse* or slipped)) ) S11. TI ( ((lumb* or spin* or vertebr*) N4 pain) ) OR AB ( ((lumb* or spin* or vertebr*) N4 pain) ) OR KW ( ((lumb* or spin* or vertebr*) N4 pain) ) S10. DE "Prognosis" S9. DE "Back Pain" S8. DE "Cohort Analysis" OR DE "Followup Studies" OR DE "Longitudinal Studies" OR DE "Prospective Studies" OR DE "Mortality Risk" OR DE "Mortality Rate" S7. S5 OR S6 S6. (DE "Sleep" OR DE "Dreaming" OR DE "Napping" OR DE "NREM Sleep" OR DE "REM Sleep" OR DE "Sleep Onset" OR DE "Sleep Quality" OR DE "Snoring" OR DE "Sleep Wake Disorders" OR DE "Hypersomnia" OR DE "Insomnia" OR DE "Narcolepsy" OR DE "Parasomnias" OR DE "Sleep Apnea") OR (DE "Bruxism" OR DE "Restless Leg Syndrome" OR DE "Sleepwalking") S5. TI ( sleep* OR parasomni* OR hyposomni* OR insomni* OR dyssomni* ) OR AB ( sleep* OR parasomni* OR hyposomni* OR insomni* OR dyssomni* ) OR KW ( sleep* OR parasomni* OR hyposomni* OR insomni* OR dyssomni* ) S4. (TI (predict* OR prognos* OR course OR cohort*) OR AB (predict* OR prognos* OR course OR cohort*) OR KW (predict* OR prognos* OR course OR cohort*)) S3. (TI "follow up stud*" or AB "follow up stud*") S2. (TI (spondylolysis OR spondylolisthesis OR spondylosis OR lumbago OR ischialgia OR dorsalgia OR "sciatic neuropathy" OR sciatica OR coccydynia) OR AB (spondylolysis OR spondylolisthesis OR spondylosis OR lumbago OR ischialgia OR dorsalgia OR "sciatic neuropathy" OR sciatica OR coccydynia) OR KW (spondylolysis OR spondylolisthesis OR spondylosis OR lumbago OR ischialgia OR dorsalgia OR "sciatic neuropathy" OR sciatica OR coccydynia)) 62 S1. (TI ("back pain*" OR backache* OR "back ache*") OR AB ("back pain*" OR backache* OR "back ache*") OR KW ("back pain*" OR backache* OR "back ache*")) 63 Supplementary Table 1. Reported and calculated effect sizes of included studies. Study ID Effect sizes for each comparison (e.g., exposure – outcome) Alsaadi 2014 Baseline sleep quality – pain intensity unadjusted: β=2.08, 95% CI: 1.99, 2.16; adjusted: β=2.00, 95% CI: 1.90, 2.09 Lovgren 2014 Effect size not used for data synthesis due to unclear final sample size Lusa 2015* Baseline sleep quality - recovery ‘mild’ poor sleep quality - unadjusted: OR=1.00, 95% CI: 0.26, 3.84 ‘severe’ poor sleep quality - unadjusted: OR=0.37, 95% CI: 0.03, 4.37 Nordeman 2017 Baseline sleep quality - disability unadjusted: r=0.16, p=0.099 Baseline sleep quantity - disability unadjusted: r=0.18, p=0.054 Kovacs 2018 Baseline sleep quality – pain intensity unadjusted: OR=0.99, 95% CI: 0.94, 1.06 Baseline sleep quality – disability unadjusted: OR=0.99, 95% CI: 0.93, 1.05 Changes in sleep quality – changes in pain intensity unadjusted: OR=4.34, 95% CI: 2.21, 8.51 Changes in sleep quality – changes in disability 64 unadjusted: OR=4.60, 95% CI: 2.29, 9.27 Pakpour 2018 Baseline sleep quality – recovery unadjusted: OR=1.52, 95% CI: 1.1