Effect of use of NSAIDs or steroids during the acute phase of pain on the incidence of chronic pain: a systematic review and meta-analysis of randomised trials

Background This study is the first to summarize the evidence on how the use of anti-inflammatory drugs during acute pain has an impact on the development of chronic pain. Methods Randomized controlled trials retrieved from nine databases included anti-inflammatory drugs (NSAIDs or steroids) versus non-anti-inflammatory drugs in patients with acute pain and reported the incidence of chronic pain. No specified date, age, sex, or language restrictions. Subgroup analyses were performed according to pain classification, follow-up time, and medication. The GRADE method was used to evaluate quality of evidence. Results A total of 29 trials (5220 patients) were included. Steroids or NSAIDs did not reduce the incidence of chronic nociceptive pain. Steroid use in acute phase significantly reduced the incidence of chronic neuropathic pain. In subgroup analysis, benefits were observed for methylprednisolone and dexamethasone, with some adverse effects. Steroids or NSAIDs were statistically significant in reducing pain intensity over 1 year, but the effect size was too small, and whether the long-term effect is clinically relevant needs to be further studied. Conclusion Quality of the evidence was low to moderate. No drug can be recommended to prevent chronic nociceptive pain. Injections of steroids (methylprednisolone or dexamethasone) during the acute phase reduce the incidence of chronic neuropathic pain, but most included studies also used local anesthetics. The results are indirect and need to be interpreted with caution. The pooled data effect sizes for pain intensity were small, so the clinical relevance was unclear. Study registration PROSPERO (CRD42022367030). Supplementary Information The online version contains supplementary material available at 10.1007/s10787-023-01405-8.


Introduction
Once the pain lasts too long and far beyond the time for the body itself to recover, it becomes a burden (Clauw et al. 2019).The impact of even very low levels of pain on social and physical functioning is significant in terms of patients' attitudes towards their own health (Voscopoulos and Lema 2010).Chronic pain affects 8.3-13% of the population in the UK (Fayaz et al. 2016;Elliott et al. 2002).In the USA, about one in three people suffer from chronic pain, more than the number of people with cancer, diabetes and heart disease combined.It costs $635 billion a year (Gaskin and Richard 2012).The Global Burden of Disease Study 2016 again pointed out that pain and pain-related conditions are the leading causes of the global burden of disability and disease (GBD 2016 Disease andInjury Incidence andPrevalence Collaborators, 2017).
Chronic pain is defined as "pain which has persisted beyond normal tissue healing time (3 months)" according to the International Association for the Study of Pain (IASP) (Pain terms 1979;Treede et al. 2019).A study by Marc Parisien (Parisien et al. 2022) found that the use of anti-inflammatory drugs (steroids or NSAIDs) hindered pain recovery during acute pain phase.This study's conclusions are groundbreaking.The results should be replicated before major changes to current clinical practice can be made.
Chronic pain is a priority in post-acute and long-term care (Drake et al. 2019).Epidemiological studies have explored the risk factors for the development of chronic pain (Mills et al. 2019).While, few interventions to prevent chronic pain have been identified (Gewandter et al. 2015).No previous study has evaluated the effect of the use of anti-inflammatory drugs during the acute phase on the incidence of chronic pain.To improve the knowledge based on existing evidence, explore how the use of anti-inflammatory drugs during acute pain affects the development of chronic pain, we designed and conducted this study.

Study eligibility
Trials were eligible for inclusion if they: (1) Public published or registered with ClinicalTrials.gov,full data available, randomized, controlled, parallel designed trials with a followup of at least 3 months [pain lasting more than 3 months was defined as chronic pain (Treede et al. 2019)].(2) The types of pain studied were nociceptive pain (including pain in the bones, muscles and skin, such as acute post-operative pain (APSP), etc. (Cohen et al. 2021)) or neuropathic pain [such as herpes zoster, acute lumbar radiculopathy, etc. (Finnerup et al. 2016)], with a duration of no more than 3 months before the first visit.(3) Anti-inflammatory drugs (NSAIDs or steroids) versus any non-anti-inflammatory control.(4) Outcome measure must include the incidence of chronic pain or pain intensity.Studies were excluded if they: (1) were animal or cell experiments, (2) were nonpharmacologic or follow-up time was less than 3 months.(3) Were cancer pain, bacterial infections, visceral pain, muscle cramps and nociplastic pain.(4) The type of pain could not be accurately resolved.This study is registered with PROSPERO (ID CRD42022367030).We anticipated that there would be great heterogeneity in the retrieved literature in terms of pain types and interventions, so we adjusted the original protocol before analysis including screening studies and subgroup analyses.

Search strategy
This study followed the PRISMA Statement.One investigator (HLY) searched 9 databases: PubMed, Web of Science, Embase, the Cochrane Library, ClinicalTrials.gov,CBMdisc, CNKI, Wanfang Database and VIP.No specified date, age, sex, or language restrictions (search strategy is shown in supplementary Table 1).We also searched cited references of relevant trial reports and reviews for potentially eligible studies.

Selection of studies
After deleting the duplicate literatures through Noteexpress (v3.9.0.9588) software, two reviewers (HLY, LG) independently screened all the study by reading the title and abstract.All studies deemed eligible after reading the title and abstract were reviewed by two pairs of reviewers (HLY, BHZ, WL, TJW) to determine whether they met the inclusion criteria.Disagreements were resolved by consensus or by consulting an adjudicator (MXH or YJZ).

Data extraction
Two paired reviewers (MYJ, JSY; HLY, LG) independently extracted this information from each study.Data extraction tables were designed to extract study characteristics and outcome information as recommended in the Cochrane Handbook (Higgins et al. 2022): year of publication, authors, journal, geographic location, funding source, design, participants, cohort size, aims and intervention, follow-up time point and outcomes.If the data of the study were represented by figures, the data were extracted by two rreviewers (HLY, LG) using data extraction software (EngaugeDigitizer 11.3).For dates not reported, the two reviewers (HLY, LG) calculated from the available data according to the conversion formula suggested in Cochrane Handbook.

Outcomes of interest
The primary outcome is the number of participants with persistent pain (three months or greater).The secondary outcomes was adverse events.

Risk of bias assessment
The risk of bias in the included studies was assessed by two researchers(MYJ, YXX) using the Cochrane Collaboration Risk of Bias Tool (Sterne et al. 2019), for risk assessment of bias in RCT.The evaluator should make a low bias risk, high bias risk and unclear judgment for each project.Disagreements were resolved by consensus or by consulting an adjudicator (XL).

Data analysis
RevMan 5.3 software was used for meta-analysis of the data.Comparisons were 2-tailed using a P < 0.05 threshold.The effect size was pooled across trials using a random (I 2 < 50%) or fixed (I 2 > 50%) effects model and heterogeneity was expressed using the I 2 statistic.For continuous data, standardized mean difference (SMD) or mean difference (MD) and 95% CI were used as the effect analysis statistic.For dichotomous data, we calculated relative risks (RR) and 95% CI on the basis of the frequency of events in each treatment group.The results of data analysis are presented using forest plots.Sensitivity analysis was performed by removing each study individually to assess the consistency and quality of the results.

Subgroup analyses
Nociceptive and neuropathic pain were divided into two subgroups for analysis.Then we divided the above two subgroups into 1-2 months, 3 months, 6 months and 1-2 years according to the different follow-up time.Finally, steroids and NSAIDs were analyzed separately.Other confounding factors affecting the analysis such as causes of pain and interventions will also be included in different subgroups for analysis.

Quality of evidence
We applied the GRADE approach to evaluate the overall quality of the evidence (Guyatt et al. 2008, b).Two independent reviewers (MXH, XL) assessed quality of the evidence from all studies and categorized it into four levels: high, moderate, low, and very low.Disagreements were resolved by consensus.

Study selection and characteristics
A total of 3803 articles were retrieved in nine databases, and 116 articles were found through other sources such as references.After removing 779 duplicate articles, the remaining 3140 articles were screened.Finally, 29 eligible articles were included (Fig. 1).Table 1 summarizes the characteristics of the studies included in the evidence synthesis.The duration of pain was less than 3 months in all participants before the first visit.

Risk of bias and grading of evidence assessment
Eighteen studies had a low risk of bias, 3 studies had an unclear risk of bias, and 8 studies had a high risk of bias (supplementary Fig. 1).Quality of the evidence was moderate overall (Table 2).The heterogeneity and imprecision of the combined effect size was the main reason for the downgrade.

The number of participants with persistent pain
Steroids Of the 9 studies involving steroids, 5 studies (1556 participants) reported the incidence of chronic pain (Turan et al. 2015;Romundstad et al. 2006;Bogefeldt et al. 2008;Saied et al. 2015;Li et al. 2018).Moderate quality evidence supports that steroid use during the acute phase of pain does not reduce the incidence of chronic nociceptive pain compared with control (RR 0.94 [95% CI 0.85-1.04],P = 0.25, I 2 = 13%; Fig. 2), with no reported adverse events.Subgroup analyses of the three studies (Turan et al. 2015;Romundstad et al. 2006;Li et al. 2018) involving postoperative pain (n = 1314) showed no difference in the effect of steroids on the incidence of chronic pain compared with the control group (RR 0.98 [95% CI 0.88-1.09];supplementary Fig. 2).Two other small sample studies (Bogefeldt et al. 2008;Saied et al. 2015) of non-postoperative pain (n = 242) showed that steroids can reduce the incidence of chronic pain, but neither reported therapeutic doses (RR 0.72 [95% CI 0.54-0.96];supplementary Fig. 3), results need to be interpreted with caution.

NSAIDs
Of the 12 studies involving NSAIDs, 9 studies (2015 participants) reported the incidence of chronic pain (Sun et al. 2013;Romundstad et al. 2006;Ling et al. 2016;Fransen et al. 2006;Bugada et al. 2015;Hancock et al. 2007;Wang et al. 2014;Haddad et al. 2019;Ge et al. 2021).Verylow quality evidence supports that NSAIDs use during the acute phase of pain does not reduce the incidence of chronic nociceptive pain compared with control (RR 0.80 [95% CI 0.64-1.00],P = 0.05, I 2 = 54%; Fig. 3).Subgroup analysis of 8 studies (n = 1776) involving postoperative pain showed a benefit of NSAIDs in reducing the incidence of chronic pain during the one-year follow-up period (RR 0.72 [95% CI 0.55-0.96];supplementary Fig. 5), although this benefit was observed only at the one-year time point (n = 304, P for interaction = 0.04).The quality of evidence is very low and the results need to be interpreted with caution.Only one (Hancock et al. 2007) of the 9 studies involved non-postoperative pain and subgroup analysis was not possible.

Intensity of pain
Steroids Of the 9 studies involving steroids, 6 studies (358 participants) reported intensity of pain (Dojode 2012; Lin  . 2010;Spijker-Huiges et al. 2014;Saied et al. 2015;Mardani-Kivi et al. 2015;Li et al. 2018).Low quality evidence supported the use of steroids to reduce pain intensity in chronic nociceptive pain over the 6-month follow-up period (SMD − 0.46 [95% CI − 0.72 to − 0.19], P = 0.0007, I 2 = 67%; Fig. 4), with no adverse effects observed.Subgroup analysis of 5 studies involving non-postoperative pain (n = 306) showed the same trend (SMD − 0.43 [95% Fig. 4 Steroids and NSAIDs and intensity of nociceptive pain CI − 0.71 to − 0.15]; supplementary Fig. 8).Only one (Li et al. 2018) of the 6 studies involved postoperative pain and subgroup analysis was not possible.There was considerable heterogeneity in the results.Sensitivity analysis showed that the heterogeneity was significantly reduced after removing the two studies (Dojode 2012;Mardani-Kivi et al. 2015).It may be caused by the significant difference in efficacy between the selected control groups (bupivacaine local injection, intermediate shock wave).

NSAIDs
Of the 12 studies involving NSAIDs, 5 studies (1270 participants) reported intensity of pain (van Helmond et al. 2016;Sun et al. 2013;Fransen et al. 2006;Comez et al. 2015;Shin et al. 2013).Moderate quality evidence supported the use of NSAIDs to reduce pain intensity in chronic nociceptive pain over the 6-month follow-up period (SMD − 0.29 [95% CI − 0.54 to − 0.04], P = 0.02, I 2 = 75%; Fig. 4), with bleeding as an adverse effect (ibuprofen 400 mg po., tid./14 days).Sensitivity analysis showed that heterogeneity was significantly reduced after excluding one studies (Fransen et al. 2006), from 75 to 61%.The other four studies were all administered by injection, and this study was administered by oral administration, which may be the cause of heterogeneity.Subgroup analysis of 4 studies involving postoperative pain (n = 1212) showed the same trend (SMD − 0.31 [95% CI − 0.59 to − 0.03]; supplementary Fig. 9).Only one (Shin et al. 2013) of the 5 studies involved non-postoperative pain and subgroup analysis was not possible.

The number of participants with persistent pain
Of the 10 studies involving steroids, 9 studies (1432 participants) reported the incidence of chronic pain (Benoldi et al. 1991;van Wijck et al. 2006;Makharita et al. 2012Makharita et al. , 2015;;Goldberg et al. 2015;Cui et al. 2017Cui et al. , 2018;;El-Sayed et al. 2021;Ni et al. 2017).Low quality evidence supports a significant reduction (53% lower) in the incidence of chronic neuropathic pain with the use of steroids in the acute phase compared with the control group during the 6-month follow-up period (RR 0.47 [95% CI 0.36-0.62],P < 0.0001, I 2 = 72%; Fig. 5).Adverse reactions included insomnia, nervousness, increased appetite, dyspepsia, dizziness, headache, arthralgia, and arthralgia.The pooled results showed great heterogeneity, and sensitivity analysis showed that the heterogeneity was reduced to 65% when one study (Goldberg et al. 2015) was removed.This study looked at the effect of oral prednisone on acute radiculopathy, whereas the other studies were all acute herpes zoster and most were administered in combination with local anesthetic injections.Subgroup analysis of 8 studies (Benoldi et al. 1991;van Wijck et al. 2006;Makharita et al. 2012Makharita et al. , 2015;;Cui et al. 2017Cui et al. , 2018;;El-Sayed et al. 2021;Ni et al. 2017) involving acute herpes zoster (n = 1163) showed that steroids significantly reduced the incidence of chronic pain (67% lower) over a 6-month follow-up period compared with usual care(RR 0.33 [95% CI 0.24-0.48];supplementary Fig. 10).

Summary of evidence
Within 29 (n = 5220) randomized controlled trials, APSP was the main research model for nociceptive pain (n = 3788) and acute herpes zoster is for neuropathic pain (n = 1553).The synthesis of available evidence suggests that neither steroids (moderate quality) nor NSAIDs (very low quality) can be recommended to prevent chronic nociceptive pain during the acute phase of pain.The use of steroids in the acute phase significantly reduced the incidence of chronic neuropathic pain (53% lower, low quality).There are insufficient data to assess the association between NSAIDs and neuropathic pain.Both steroids (low quality) and NSAIDs (moderate quality) reduce pain intensity, but the combined effect sizes are small and there is substantial heterogeneity.Whether the long-term effect is clinically relevant needs to be further studied.
Subgroup analyses by cause of pain (postoperative or non-postoperative pain; acute herpes zoster or acute radiculopathy) showed the same trend as described above.Subgroup analyses showed that methylprednisolone (40 or 80 mg) and dexamethasone (8 mg) combined with local anesthesia injections significantly reduced the incidence and intensity of chronic neuropathic pain, with adverse effects including dizziness, headache, back pain, and somnolence.Administration methods include epidural injection, intradermal injection and paravertebral injection.Neither methylprednisolone (125 mg or 250 mg iv.), parecoxib (40mgiv.),nor ketorolac (30 mg iv./2 mg intradural injection) reduced the incidence of chronic nociceptive pain.Due to limitations in the number and design of the studies, we were unable to do separate subgroup analyses for all drugs.

Limitations of previous studies
Previous studies have discussed the relationship between anti-inflammatory drugs and chronic nociceptive pain, mainly using postoperative pain as a research model.The study by Ian Gilron et al. (Carley et al. 2021), which reviewed the evidence for the use of drugs to prevent chronic nociceptive pain after surgery, showed that none of the drugs studied to date could be recommended.Compared with similar degrees of nociceptive pain, neuropathic pain might be associated with greater decrements in quality of life (Saavedra-Hernández et al. 2012;Spahr et al. 2017).However, there is little evidence summarizing the incidence of chronic neuropathic pain.Many epidemiological studies have explored the risk factors for the development of chronic pain, while some factors can identify the individuals who are at risk of developing chronic pain, few interventions to prevent chronic pain have been identified (Gewandter et al. 2015).

Strengths and limitations of the study
There are some limitations to this meta-analysis.Twelve different drugs were used in the included studies, including five steroids and seven NSAIDs, and some were administered in different ways and for different durations.This leads to heterogeneity in partially integrated data, which is one of the main reasons for lowering the level of evidence.Studies of injectable administration in the treatment of neuropathic pain have all combined local anesthetics, making it difficult to assess the effect of steroids on the incidence of chronic neuropathic pain.We reduced the level of evidence due to indirectness.The incidence of chronic pain was not a primary outcome in some studies, which may be at risk for selective reporting.For pain intensity, the effect size in the pooled data was small and, although statistically significant, may not be clinically relevant.The small number of disease types included in the study is also a limitation of this study, with nociceptive pain mainly APSP and neuropathic pain mainly acute herpes zoster.Whether the conclusions can be generalized to pain caused by other diseases needs to be cautious.Some studies did not report the potential safety issues of the drug intervention, which was an impediment to conducting quantitative assessments to weigh the benefit-risk trade-offs.In the process of disease development, the patient's psychological factors, age, gender, etc. can also affect pain, but the current research has no way to evaluate or effectively eliminate the influence of these confounding factors.
This study focuses on how the use of anti-inflammatory drugs during the acute phase of pain affects the incidence of chronic pain and has several strengths.(1) This is the most recent review of the effects of acute-phase anti-inflammatory drug therapy on the incidence of chronic pain; (2) we conducted a comprehensive search of eligible RCTs in all languages, with no specified date, age, sex, or language restrictions, and also searched citations of relevant trial reports and reviews for potentially eligible studies; (3) all factors that may affect the results were considered in the analysis of the data.Subgroup analyses were designed according to the characteristics of the available studies, including pain type, follow-up time, drug, dose, and duration and (4) this review follows the PRISMA statement, and the procedures throughout the review process are rigorous and reproducible.

Interpreting the findings
For neuropathic pain, analgesic antidepressants and antiepileptic drugs are first-line medications based on many placebo-controlled trials that are of moderate and high quality (Finnerup et al. 2015;Derry et al. 2013Derry et al. , 2014)).However, there is still no consensus on the effectiveness of the prevention of chronic neuropathic pain.The current data show that the combination of anti-inflammatory drugs (methylprednisolone and dexamethasone) in the acute phase of neuropathic pain may provide greater benefits to patients.But most steroids are combined with a local anesthetic, making the results indirect.Specifically designed single-drug intervention RCTs with long follow-up are needed.Effects of inflammatory response and other factors can last for a period of time (days or weeks), and studies of a single-shot drug intervention may be difficult to find valuable results.

Conclusion
The quality of evidence was low to moderate.Steroids or NSAIDs have analgesic effects in the acute phase but do not reduce the incidence of chronic nociceptive pain.For neuropathic pain, steroids use in the acute phase significantly reduced the incidence of chronic pain at 3-6 months (by 53%).Only methylprednisolone and dexamethasone had an observed benefit, with some adverse effects.However, the evidence is indirect and needs to be interpreted with caution.

Fig. 1
Fig. 1 Flow diagram of study selection

Fig. 3
Fig. 3 NSAIDs and incidence of chronic nociceptive pain