Background

Low back pain remains the leading cause of years lived with disability (YLD) worldwide [1]. In 2017, low back pain was responsible for around 64.9 million YLD, an increase of 47.5% since 1990; YLD due to neck pain also increased by 65.8% during the same period [1]. In North America, opioids are commonly prescribed to relieve low back and neck pain [2]; however, opioids provide only modest improvements in pain intensity, physical function, and sleep quality [3, 4]. Moreover, opioids are associated with rare but catastrophic risks, including nonfatal and fatal unintentional overdose, and 1 in 20 patients prescribed opioids for chronic pain will develop an opioid use disorder [5,6,7]. Accordingly, current clinical practice guidelines recommend optimization of non-opioid pharmacotherapy and non-pharmacologic treatments (e.g. education, exercise, cognitive behavioural therapy, soft-tissue massage, spinal manipulation) rather than prescription opioids as first-line therapy for acute or chronic non-cancer musculoskeletal pain [6, 8, 9].

In several recent studies, receipt of chiropractic care has been associated with lower chances of receiving prescription opioids [10,11,12,13,14,15,16] and, among those already prescribed, reduced opioid dose [17,18,19] among patients with non-cancer spine pain. A 2022 observational study of 40,929 opioid-naive persons with new-onset low back pain found that those who received chiropractic treatment early in their complaint had 12% lower odds of incident opioid use (adjusted odds ratio [OR] = 0.88; 95% confidence interval [CI], 0.80 to 0.97) and 44% lower odds of long-term opioid use (adjusted OR = 0.56; 95% CI, 0.40 to 0.77) [13]. A 2022 mixed-methods analysis of 210 opioid-using community health centre patients with chronic back or neck pain [19] found that the rate of prescription opioid fills over 12-month follow-up was 34% lower (adjusted incidence rate ratio (IRR) = 0.66; 95% CI, 0.52 to 0.83), and refills were 73% lower (adjusted IRR = 0.27; 95% CI, 0.17 to 0.42), among those who initiated chiropractic care (n = 49) versus non-recipients (n = 161). Recipients were also between 78 and 86% less likely to be prescribed a higher (i.e. \(\ge\) 50-mg morphine equivalents daily [MED]) opioid dose [19]. These findings, combined with those of other reports [10,11,12,13,14,15,16,17,18,19,20], suggest that access to chiropractic services may reduce reliance on opioids for acute or chronic musculoskeletal pain.

A prior systematic review and meta-analysis of six uncontrolled observational studies [20] found an inverse association between attending a chiropractor and opioid receipt among patients with spinal pain (pooled OR = 0.36; 95% CI, 0.30 to 0.43). However, the literature search informing this systematic review was conducted up to April 18, 2018, the certainty of evidence was not examined, and assessments of risk of bias and heterogeneity were suboptimal [20]. Moreover, at least 10 additional studies investigating the effect of chiropractic care on new and existing opioid use have since been published [10,11,12,13,14,15,16,17,18,19]. As such, an updated systematic review and meta-analysis on chiropractic use and opioid receipt among patients with spinal pain is warranted.

Objectives

The purpose of our systematic review is to assess the impact of chiropractic care on (1) initiation, or continued use, of prescription opioids and (2) patient-important outcomes, including pain intensity, physical and emotional functioning, sleep quality, patient satisfaction, and adverse events, among adult patients with non-cancer spine pain. Our focus is on non-cancer spine pain because cancer-related spine pain is a contraindication to high-velocity, low-amplitude spinal manipulation [21]. We will explore whether our results are influenced by factors such as the year the study was conducted, methodological quality, whether pain is acute or chronic, type of opioid prescriber(s), earlier versus later chiropractic exposure, or frequency of chiropractic treatment visits.

Methods

We have reported our systematic review protocol in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols (PRISMA-P) statement [22] (Additional file 1). Our protocol is also registered in the International Prospective Register of Systematic Reviews (PROSPERO) with the registration number CRD42023432277.

Eligibility criteria

Our eligibility criteria are summarized below using the PICOS (i.e. Population, Intervention, Comparison intervention, Outcome measures, Study designs) framework [23].

Participants/population

Inclusion

We will include adult patients (\(\ge\) 18 years of age) with non-cancer back or neck pain (with or without radicular symptoms) of any duration.

Exclusion

We will exclude patients with spinal neoplasms or other contraindications to chiropractic treatment (i.e. ‘red flag’ diagnoses such as fractures, infections, inflammatory arthritis, or cauda equina syndrome) [21].

Intervention/exposure

Our exposure of interest will be receipt of chiropractic care, which is defined as care provided by a chiropractor, including, but not limited to spinal manipulation, soft-tissue therapy, education, reassurance, and self-care advice (e.g. icing, stretching, and strengthening exercises) [24].

Comparator/control

The comparison will be nonreceipt of chiropractic care (e.g. usual medical care, physiotherapy).

Outcomes

Primary outcomes

Our primary outcomes will be as follows: (1) prescription opioid receipt and (2) continued prescription opioid use (measured as the number and/or dose of opioid prescriptions).

Secondary outcomes

We will extract data on all other patient-important outcomes [25] that are reported, including the following: (1) pain intensity, (2) physical functioning, (3) emotional functioning, (4) sleep quality, (5) patient satisfaction, and (6) adverse events [26, 27]. We will consider outcomes for physical and emotional functioning that are reported over a minimum 4-week follow-up period.

Study designs

Inclusion

We will include both randomized and non-randomized (quasi-experimental) controlled trials and observational studies (including cohort and case–control studies) that reported an adjusted analysis exploring the association between receipt of chiropractic care and opioid use.

Exclusion

We will exclude case reports, case series, cross-sectional studies, protocols, letters, editorials, commentaries, books and book chapters, dissertations, conference abstracts, and secondary sources of evidence, including clinical practice guidelines and systematic, scoping, or narrative reviews.

We will not exclude studies based on geographic location, language, or date of publication. Studies that either were published before, or were included in, the 2020 systematic review by Corcoran et al. [20] will be re-examined to ensure a comprehensive literature search and to validate eligibility.

Information sources

We will search MEDLINE, Embase, AMED, CINAHL, Web of Science, and the Index to Chiropractic Literature without geographic or language restrictions from the inception of each database to June 2024. Our database-specific search strategies have been developed by an academic librarian (RJC) and were reviewed by a second librarian using the peer review of electronic search strategies (PRESS) checklist [28] (Additional files 2 and 3). We will also hand-search reference lists of eligible articles and related systematic reviews and contact content experts to identify additional eligible studies. We will re-run our database searches prior to our final analyses and retrieve any additional eligible studies for inclusion.

Study selection

After duplicate records have been removed (see Fig. 1), pairs of reviewers will independently screen titles, abstracts, and full-text studies for eligibility using online systematic review software (DistillerSR, Evidence Partners, Ottawa, Canada; https://www.distillersr.com/). Calibration exercises will be conducted prior to title/abstract and full-text screening to improve consistency between reviewers throughout the study selection process. Disagreements on eligibility will be resolved through discussion to achieve consensus or, when not possible, adjudication by a third reviewer. We will calculate inter-rater agreement on title/abstract and full-text screening using an adjusted kappa (\(\kappa\)) statistic [29] and interpret the strength of agreement at each stage as follows: poor (\(\kappa\) \(\le\) 0.2), fair (0.21 \(\le\) \(\kappa\) \(\le\) 0.4), moderate (0.41 \(\le\) \(\kappa\) \(\le\) 0.6), substantial (0.61 \(\le\) \(\kappa\) \(\le\) 0.8), or almost perfect (\(\kappa\) > 0.8). We will adapt our screening processes (e.g. re-train or substitute raters) if \(\kappa\) agreement on title/abstract or full-text screening is < 0.6 [30]. A PRISMA flow diagram [31] of our study selection process is provided in Fig. 1.

Fig. 1
figure 1

PRISMA flow diagram

Full size image

Data collection process

Using standardized, pre-piloted data extraction forms, pairs of reviewers will independently extract data from included studies. We will conduct calibration exercises prior to our formal data extraction and quality assessment procedures to ensure consistency between reviewers. Extracted information will include the following: (1) the last name of first author, (2) year of publication; (3) country where the study was conducted; (4) study design; (5) number of participants; (6) participant demographics (i.e. age, sex, primary pain complaint); (7) chiropractic care and control group information (e.g. proportion of patients receiving chiropractic or usual medical care; type of usual medical care provided, such as primary or specialist care; number of days between the index visit date and initiation of chiropractic care; number of chiropractic treatment sessions attended over follow-up); (8) duration of follow-up; (9) details on opioid use (i.e. proportion of sample prescribed opioids and, when available, total number and dose of opioid prescriptions); and (10) all patient-important outcomes including pain intensity, physical and emotional functioning, sleep quality, patient satisfaction, and adverse events. Discrepancies between reviewers will be resolved as previously described. We will also contact study authors when necessary to request unpublished or missing data or for clarification regarding eligibility.

Risk of bias in individual studies

Two reviewers will independently assess risk of bias of eligible randomized controlled trials (RCTs) and quasi-experimental studies using a risk-of-bias tool developed by the CLARITY group (https://www.distillersr.com/resources), according to the following domains: sequence generation; allocation concealment; blinding of patients, health care providers, data collectors, outcome assessors, and data analysts; infrequent missing data (> 20% will be considered high risk of bias); and selective outcome reporting. For this final item, we will search clinical trial registries (e.g. clinicaltrials.gov) to compare studies’ pre-specified outcomes with their published results. When protocols are not available, we will compare the methods and results in each trial publication. Response options for each item will be categorized as ‘definitely or probably yes’ (assigned as low risk of bias) and ‘definitely or probably no’ (assigned as high risk of bias). We will also use criteria suggested by the CLARITY group to assess the risk of bias of observational (i.e. cohort and case control) studies, including the following: selection bias, assessment of exposure, validity of outcome assessment(s), control of confounding variables (with adjustment for age, sex, and severity or duration of non-cancer spine pain, at a minimum, considered as an adequately adjusted model), and loss to follow-up. Disagreements between reviewers will be resolved by consensus or adjudication by a third reviewer.

Data synthesis

We will pool all binary outcomes that are reported by more than one study (e.g. prescribed opioid receipt, long-term opioid use, higher versus lower opioid dose) using ORs and associated 95% CIs. We will use a threshold of 50-mg MED to define higher versus lower opioid dose [6]. When studies provide hazard ratios (HRs) and relative risks (RRs) for time-to-event data (e.g. time-to-first opioid prescription) or incidence rates (number of opioid prescriptions over time), we will convert the HR or RR to an OR using a baseline risk (i.e. proportion of patients in the non-chiropractic care control group who had the events) before pooling [32]. Continuous outcomes such as pain intensity, physical and emotional functioning, sleep quality, or morphine equivalent dose will be pooled as weighted mean differences (WMDs) with associated 95% CIs after converting different instruments that report on the same domain (e.g. pain) into the most commonly reported scale among studies eligible for review [33, 34]. For all outcomes, we will conduct separate analyses for RCT/quasi-experimental and observational studies and prioritize adjusted over unadjusted effect estimates from observational studies if both sets of data are available.

We will conduct all meta-analyses using random-effects models [35] and the DerSimonian-Laird method [36]. We will also explore the consistency of association between our pooled results and studies reporting the same outcome domains that were unable to be pooled.

To avoid overestimating the magnitude of effect or association when restricting statistical pooling to estimates that appear in adjusted regression models, we will impute an OR of ‘1’ or WMD of ‘0’ for effects (from RCTs) and associations (from observational studies) that were tested in bivariable analyses but because of non-significance were excluded from adjusted analyses or were included in multivariable analyses with the only information provided being that they were ‘not significant’. We will impute an associated variance for all such estimates using the hot deck approach [37].

If there are ≥ 10 studies available for meta-analysis [35], publication bias will be assessed for each outcome by visual assessment of funnel plots for asymmetry and calculation of Egger’s test [38] for continuous outcomes and Harbord’s test [39] for binary outcomes. We will evaluate the certainty of evidence for all pooled measures of association using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach [40,41,42] (Tables 1 and 2). All analyses will be performed using Stata V.18 (StataCorp, College Station, TX, USA), and comparisons will be two-tailed using a statistical significance threshold (\(\alpha\)) of 5%.

Table 1 GRADE evidence profile of the associations between chiropractic care and initiation, or continued use, of prescription opioids and other patient-important outcomes for non-cancer spine pain
Full size table
Table 2 Quality of evidence levels for GRADE [40]
Full size table

Subgroup, meta-regression, and sensitivity analyses

Heterogeneity will be examined through tau-squared and visual inspection of forest plots [42, 43]. When there are at least two studies in each subgroup, we will explore sources of heterogeneity with six prespecified subgroup hypotheses, assuming larger effects or associations with the following: (1) studies conducted in earlier versus later calendar years — a proxy for increased pressure on physicians to reduce opioid prescribing [6, 8]; (2) higher versus lower risk of bias, evaluated on a criterion-by-criterion basis; (3) acute versus chronic pain; (4) general practitioner versus specialist (e.g. physiatrist/pain physician) or emergency department opioid prescriber(s) [11, 13]; (5) early versus later chiropractic exposure; and (6) lower versus higher frequency of chiropractic treatment visits [14, 19, 44]. In line with previous literature [12,13,14, 16], we will define ‘early’ chiropractic exposure as receipt of chiropractic services within the first 30 days after an index visit for acute or chronic non-cancer spine pain.

When there are at least 10 studies available [35, 45], we will use meta-regression to explore the relationship between time period or chiropractic visit frequency and the association of chiropractic care on opioid use. If we find a significant slope, we will use the distribution of the scatter plot to determine an appropriate cut-off value for our subgroup analyses involving studies conducted in earlier versus later calendar years and lower versus higher frequency of chiropractic treatment visits. Tests for interaction will be performed to establish whether subgroups differ significantly from one another, and we will assess the credibility of significant subgroup effects (test for interaction p < 0.05) using the Instrument for Assessing the Credibility of Effect Modification Analyses (ICEMAN) criteria [46].

We will conduct sensitivity analyses to examine the impact of converting ORs from HRs or RRs, and to examine the effect of imputing data for nonsignificant effects or associations.

Patient and public involvement

We did not engage patients or the public as direct contributors to the current protocol. However, we prefaced our review with two mixed-methods studies, including interviews of people with lived and living experience of spine pain, opioid use, and chiropractic care [14, 19] that helped inform the current project. Further, we relied on the Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials [25, 47], which was informed by patient focus groups and surveys [48], to select patient important outcomes.

Ethical considerations

This is a systematic literature review of previously published studies and does not require ethics approval.

Discussion

Opioid-related morbidity and mortality have risen in several countries over the past 25 years but particularly in Canada and the United States (US) [49]. In Canada, there were 44,592 opioid-related deaths and 42,711 opioid-related hospitalizations between January 2016 and December 2023 [50]. In the USA, there were more than 70,000 opioid-related deaths in 2020 alone [49]. Young- to middle-aged adult men have been most affected by the opioid crisis [49,50,51], which has arisen partly among individuals who were initially prescribed opioids for back pain or some other musculoskeletal condition [49, 51,52,53]. The US Centers for Disease Control and Prevention estimated the annual cost of the opioid crisis at over US $1 trillion in 2017, equivalent to 5% of US gross domestic product [49, 54]. Recent reports from Canada and the USA also indicate that rates of opioid-related deaths and hospitalizations have worsened since before the COVID-19 pandemic [49, 50, 55].

The ongoing opioid crisis in North America has generated interest in exploring treatment options that may reduce reliance on opioids for patients with spine-related or other musculoskeletal pain. Findings from a 2020 systematic review [20] and some subsequent primary studies suggest that utilization of chiropractic services may be effective in reducing opioid prescribing [10,11,12,13,14,15,16, 20] and long-term opioid use [17,18,19]; however, the overall magnitude and certainty of these effects are unknown. The aim of our systematic review will be to assess the impact of chiropractic care on initiation, or continued use, of prescription opioids for adult patients with non-cancer spine pain. The results of our systematic review will be of interest to patients, health care providers, and policy-makers.

Strengths and limitations

Our review has several strengths. First, we will use explicit eligibility criteria and conduct a comprehensive search without date, geographic, or language restrictions to identify RCTs and observational studies exploring the impact of chiropractic care for spine pain and opioid receipt [20]. Second, we will assess the risk of bias among individual studies and evaluate the certainty of evidence using the GRADE approach. Third, we will use pre-defined subgroup analyses to explore sources of heterogeneity, and we will assess the credibility of all potential subgroup effects. Fourth, we will conduct sensitivity analyses to confirm the robustness of our meta-analyses. A limitation of our review is that we anticipate most eligible studies will be observational, which may limit the strength of inferences from our results.

Knowledge translation

The results of our review will be disseminated via a peer-reviewed publication and conference presentations.