Topical diclofenac versus oral ibuprofen versus diclofenac + ibuprofen for ED patients with acute low back pain: A randomized study

Abstract

Introduction.

Topical non-steroidal anti-inflammatory drugs (NSAIDs) are useful for a variety of musculoskeletal injuries. It is not known whether topical NSAIDs should be used for patients presenting with acute non-radicular musculoskeletal low back pain (LBP).

Methods.

We conducted a randomized, placebo-controlled double-blind study in which ED patients 18-69 years with acute, non-traumatic, non-radicular, musculoskeletal LBP were randomized at the time of discharge to treatment with oral ibuprofen 400mg orally + placebo topical gel, 1% diclofenac topical gel + oral placebo, or ibuprofen 400mg + 1% diclofenac topical gel. We measured outcomes using the Roland Morris Disability Questionnaire (RMDQ), a 24-item yes/no instrument about the impact of back pain on a respondent’s daily activities. The primary outcome was change in RMDQ between ED discharge and two days later. Medication related adverse events were elicited by asking whether the study medications caused any new symptoms.

Results.

3281 patients were screened for participation and 198 were randomized. Overall, 36% of the population were women, the mean age was 40 years (SD 13), and the median RMDQ score at baseline was 18 (25^th, 75^th percentile: 13, 22) indicating substantial low back related functional impairment. 183 (92%) participants provided primary outcome data. By two days after the ED visit, the ibuprofen + placebo group had improved by 10.1(95%CI 7.5 to 12.7), the diclofenac gel + placebo group by 6.4(95%CI 4.0 to 8.8), and the ibuprofen + diclofenac gel by 8.7(95%CI 6.3 to 11.1). The 95%CI for between group differences were as follows-- ibuprofen versus diclofenac: 3.7 (0.2, 7.2); ibuprofen versus both medications 1.4 (−2.1, 4.9); diclofenac versus both medications −2.3 (−5.7, 1.0). Medication related adverse events were reported by 3/60 (5%) ibuprofen patients, 1/63 (2%) diclofenac patients, and 4/64 (6%) patients who got both.

Conclusions.

Among patients with non-traumatic, non-radicular acute musculoskeletal low back pain discharged from an ED, topical diclofenac was probably less efficacious than oral ibuprofen. It demonstrated no additive benefit when co-administered with oral ibuprofen.

INTRODUCTION

Background

In industrialized nations such as the United States, low back pain carries a lifetime prevalence of over 80% and results in an aggregate 2.7 million annual visits to US emergency departments. (1,2) Randomized controlled trials comparing NSAIDS to placebo support the use of NSAIDs as first-line treatment in patients with acute low back pain without sciatica. (3) However, the magnitude of NSAID effectiveness remains small, resulting in many patients with low back pain reporting persistent pain, analgesic use, and functional impairment after an emergency department (ED) discharge. (3,4) The addition of other pharmacologic treatments such as oxycodone, muscle relaxants, and benzodiazepines have not improved outcomes when added to NSAIDS. (5,6,7,8)

Importance

Topical NSAIDs can be used for musculoskeletal pain with the goal of delivering medications directly to injured tissue while minimizing systemic side effects. Topical NSAIDS provide similar concentrations of the drugs in muscle tissue but much lower plasma concentration than oral formulations. (9) Commercially available topical NSAID formulations include diclofenac, ibuprofen, and ketoprofen, although only the former is available in the US. (10) Among patients with musculoskeletal sports injuries, topical NSAIDs improve pain more than placebo. (11, 12) While, there is no high quality evidence supporting the use of these medications for acute musculoskeletal low back pain, lower quality data suggest benefit. (13, 14)

Goals of this investigation

Given the poor pain and functional outcomes that persist beyond an ED visit for musculoskeletal low back pain and the need for effective treatments, we conducted a randomized, double-blind trial to evaluate the role of diclofenac gel, a topical NSAID, in the treatment of acute, non-traumatic, non-radicular low back pain. Specifically, we examined its potential benefit as both a synergistic drug and as a replacement for oral NSAIDS by testing the following hypotheses:

1. Combining topical diclofenac gel with oral ibuprofen will result in better LBP functional outcomes than either drug alone as measured by improvement in the Roland Morris Disability Questionnaire (RMDQ) two days after ED discharge.

2. Oral ibuprofen will result in better LBP functional outcomes than topical diclofenac, as measured by improvement in the Roland Morris Disability Questionnaire 2 days after ED discharge.

MATERIALS AND METHODS

Study Design and Setting

This was a randomized, double-blind, placebo-controlled comparative effectiveness trial in which we enrolled patients during an ED visit for musculoskeletal LBP and followed them by telephone two and seven days later. The Albert Einstein College of Medicine Institutional Review Board reviewed and approved this study. Written consent was obtained from all study participants. This study was registered online at http://clinicaltrials.gov(NCT04611529). Enrollment commenced in March 2021 and concluded in May 2023. This study was conducted in 2 EDs with a combined volume of 175,000 annual visits. Both EDs are affiliated with Montefiore Medical Center, an urban teaching hospital system. Salaried, trained, fluently bilingual (English and Spanish) research associates staffed the ED 24 hours per day, 7 days per week during the accrual period. Our study adhered to the CONSORT guidelines for randomized trials. (15)

Selection of Participants

Our goal in this study was to include a broad representation of patients with musculoskeletal back pain who were likely to respond to the investigational medications and who would not be considered candidates for spinal surgery or targeted epidural intervention. We hoped for a widely generalizable study and therefore did not require diagnoses to be contingent on advanced imaging studies. We included adults aged 18-69 who presented to the ED primarily for low back pain, defined as pain originating between the lower border of the scapula and the upper gluteal folds. Patients were considered for inclusion if the primary clinical diagnosis was that of non-traumatic, non-radicular, musculoskeletal pain and if they were discharged home at the conclusion of the ED visit. Patients were included only if their back pain was deemed functionally impairing, which we defined as a score of greater than 5 on the Roland-Morris Disability Questionnaire (RMDQ). (16) The RMDQ is a 24-item questionnaire commonly used to measure LBP and related functional impairment (0 indicates no impairment; 24 indicates maximum impairment) (https://www.rmdq.org/). (4, 5, 6, 7, 8, 16) We excluded patients presenting with radicular pain, which we defined as pain radiating in a dermatomal pattern below the gluteal folds, pain of traumatic etiology, and pain lasting longer than 2 weeks or occurring at baseline more frequently than once per month. We also excluded patients who were pregnant, had open wounds of the lower back, used daily or near daily pain medications prior to the onset of acute low back pain, and those with contra-indications or allergies to the investigational medications.

Interventions

The pharmacist performed randomization in blocks of 6 based on a sequence generated at http://randomization.com. Patients were randomized in a 1:1:1 manner to one of 3 interventions: the ibuprofen arm (ibuprofen 400mg orally plus placebo gel every 6 hours as needed), the diclofenac arm (diclofenac 1% gel, 4g topically, plus placebo capsules every 6 hours as needed), and the combination therapy arm (ibuprofen 400mg orally plus diclofenac 1% gel, 4g topically, every 6 hours as needed). We chose 400mg as the dose for ibuprofen as benefit with larger doses is unlikely . (17) We dispensed a two-day supply of investigational medication to each patient.

An investigational pharmacist masked the investigational oral medication by placing tablets containing active medication or nothing at all into identical capsules, which were packed with scant amounts of lactose and sealed. The investigational topical medication was masked by placing active or identically appearing inactive gel into an unmarked tube. This masking took place in a secure location inaccessible to ED personnel. Thus, all clinical staff, study participants, outcome assessors and researchers were blinded to treatment. The investigational pharmacist, who was not in the ED, was not blinded. The clinical nurse presented each study participant with a two-day supply of investigational medications as one bottle of capsules and one tube of gel along with dosing instructions. Additionally, they were provided with a measuring tape and instructed to measure a 4.5-inch length of gel corresponding to a 4gm dose of the investigational topical medication for each application. Patients were instructed to always use both the oral and topical medication together on an as needed basis for pain control and cautioned not to take off-protocol medication for LBP without first consulting with a healthcare provider.

Prior to discharge, all patients were provided with a 10-minute educational intervention by the research associate. This was based on NIAMS’s Handout on Health: Back Pain information webpage (available at https://www.niams.nih.gov/health-topics/back-pain/basics/diagnosis-treatment-and-steps-to-take)

Outcome measures

The primary outcome for this study was the change in Roland Morris Disability score between the baseline ED visit and the two-day follow-up. A 5-point improvement on this scale is generally considered clinically significant. (16) As a secondary outcome, we also evaluated the change in Roland Morris Disability score between baseline and the seven-day follow up. Other secondary outcomes included the participants’ worst pain during the 24 hours prior to the two follow-up calls and whether the study participant used any pain medication during the 24 hours prior to the follow-up calls. We ascertained both of these outcomes at both the two-day and seven-day follow-up. We measured participants’ worst pain using a 4-point ordinal scale (none, mild, moderate, or severe). Finally, during the scheduled follow-up, we asked participants whether they experienced any new symptoms related to the study medications and followed up affirmative responses with an open-ended query eliciting details.

The research associates collected all data through interviews with the study participants. We collected baseline data through in-person interviews in a quiet area of the ED. All follow-up data were collected by the RAs through telephone interviews.

Analysis

We report baseline characteristics as mean (SD), median (25^th, 75^th percentile) or number (percent), as appropriate. We performed an intention-to-treat analysis among all patients for whom outcome data were available. We made no assumptions about patients lost-to-follow-up.

The primary outcome was a comparison of the Roland-Morris Disability Questionnaire score between the ED visit and 2 days later. We report the improvement in each group as the mean change with 95% confidence intervals (CI). The between-groups differences were calculated using the mean RMDQ change of each group and reported as mean differences with 95%CI. We considered differences statistically significant if the 95%CI did not cross zero.

We treated seven-day RMDQ data identically, again reporting the change from baseline and the between-groups differences. Dichotomous secondary outcomes are reported as n/N (%). For analysis of the ordinal pain intensity data (severe, moderate, mild, or none), as has been done before, we lumped together moderate pain with severe and mild pain with none. We also report between group differences with 95%CI for the ordinal data and again considered differences statistically significant if the 95%CI did not cross zero.

We examined the association of baseline variables with the primary outcome graphically, visually looking for evidence of an association. All variables with graphical evidence of an association, and all variables with substantial between-group discrepancies were included in a multivariable linear regression model in which the primary outcome was the dependent variable and study meds were included as an independent variable. We built three such models for each pair of comparisons. For each model, we entered and maintained all variables in the model, report B coefficient with 95%CI and R squared for the models. We evaluate for homoscedasticity using the White and Breusch-Pagan test. For each of these, we reject the null hypothesis of not heteroscedastic if p<0.05. We determine influential cases using Cook’s distance and examined these cases for themes that might shed light on the model.

Based on data previously collected among patients with LBP who received NSAIDs alone, we expected a mean RMDQ change of 10.2 and a standard deviation of 8.9 (18). A widely accepted minimum clinically important improvement of 5 points on the RMDQ would require those randomized to the combination of medications to demonstrate a mean improvement of 15.2 on the RMDQ. Using a standard alpha of 0.05 and a beta of 0.20, we determined the need for 60 subjects in each arm. To account for protocol violations and patients lost-to-follow-up, we intended to enroll 66 patients in each arm (total n= 198).

We used SPSSv.25 (IBM Corp. Armonk, NY) for all statistical analyses.

RESULTS

Characteristics of Study Subjects

During the enrollment period, we screened 3281 patients and enrolled 198 eligible patients (See CONSORT flow diagram, Figure 1). A larger percentage of women received ibuprofen alone. Baseline characteristics were otherwise similar among the 3 groups (table 1). Most patients presented with LBP of less than 3 days duration and had an infrequent history of LBP prior to the ED visit.

Figure 1:

CONSORT flow diagram

Table 1:

Baseline characteristics

Variable	Ibuprofen+ placebo gel	diclofenac gel+ placebo capsule	ibuprofen+ diclofenac gel
	n=66	n=66	n=65
Age mean (SD), y	37 (13)	39 (13)	39 (13)
Sex
Men	32 (48)	48 (73)	45 (69)
Women	34 (52)	18 (27)	20 (31)
Race
Asian	0 (0%)	2 (3%)	0 (0%)
Black/African American	16 (25%)	19 (30%)	9 (14%)
Hispanic/Latino	43 (68%)	42 (66%)	51 (80%)
White	2 (3%)	1 (2%)	3 (5%)
Multiracial	1 (2%)	0 (0%)	0 (0%)
Other	1 (2%)	0 (0%)	1 (2%)
Refused	3	2	2
Median RMDQ at ED visit (IQR)	19 (14, 22)	17 (13, 22)	19 (13, 21)
Median duration of LBP before presentation to ED, h (IQR)	48 (24, 98)	48 (24, 96)	72 (24, 120)
Previous episodes of LBP
Never before	24 (36)	18 (27)	15 (23)
Few times before	29 (44)	37 (56)	36 (55)
once/year	12 (18)	9 (14)	14 (22)
once/month	1 (2)	2 (3)	0

Main results

At the 2 day follow-up, all 3 groups showed an improvement in the mean RMDQ score compared to baseline (table 2). Participants had a mean RMDQ improvement of 10.1 (95%CI 7.5 to 12.7) in the ibupofen group, 6.4 (95%CI 4.0 to 8.8) in the diclofenac gel group, and 8.7 (95%CI 6.3 to 11.1) in the ibuprofen + diclofenac gel group. An analysis of the mean differences between groups favored the use of ibuprofen over diclofenac (mean difference of 3.7; 95%CI 0.2 to 7.2). There were no additional differences between groups for the change in the RMDQ score. The distribution of the primary outcome is depicted graphically in Appendix Figure 2. Please see a graphical representation of the association between baseline variables and the primary outcome in Appendix Figure 1. In a multivariable model that included sex and baseline RMDQ score, ibuprofen outperformed diclofenac gel by 3.50 (0.20, 6.80) (Appendix Table 1).

Table 2:

Two-day outcomes

Variable	ibuprofen+ placebo gel	diclofen ac gel+ placebo capsule	ibuprofen+ diclofenac gel		ibuprofen vs diclofenac gel	ibuprofen vs both	diclofenac vs both
	n=66	n=66	n=65		Mean difference, % (95% CI)
Mean RMDQ change (95% CI)	10.1 (7.5 to 12.7)a	6.4 (4.0 to 8.8)a	8.7 (6.3 to 11.1)a		3.7 (0.2 to 7.2)b	1.4 (−2.1 to 4.9)b	−2.3 (−5.7 to 1.0)b
Worst LBP during previous 24h c
None	11 (18)	8 (13)	5 (8)
Mild	28 (47)	21 (33)	27 (43)		19% (2 to 36%)d	14% (−3 to 31%)d	5% (−13 to 22%)d
Moderate	14 (23)	23 (37)	18 (29)
Severe	7 (12)	11 (17)	13 (21)
Missing	6	3	2
Use of any medication for LBP during previous 24h e
Yes	53 (88)	50 (81)	54 (86)		8% (−5 to 20%)	3% (−9 to 14%)	5% (−8 to 18%)
No	7 (12)	12 (19)	9 (14)
Missing	6	4	2

Abbreviations: RMDQ- Roland-Morris Disability Questionnaire, LBP- Low back pain

Data presented as No. (%) unless otherwise stated.

^aMean RMDQ improvement for patients in each group. The RMDQ improvement for each patient calculated as (baseline RMDQ score – RMDQ score at 2-day follow up)

^bThe between-group difference of the Mean RMDQ change at the 2-day follow up

^cWe asked patients “How would you describe your worst low back pain in the last 24 hours?”

^dNone/mild versus moderate/severe

^eWe asked patients “During the last 24 hours, have you taken any medication at all for low back pain?”

We report secondary outcomes measured two days after the ED visit in Table 2. Overall, moderate or severe pain was reported by nearly 50% of the cohort and more than 80% required mediction for their back pain. Patients using only diclofenac gel were more likely to report moderate or severe pain than those using only ibuprofen (between group difference: 19%; 95% CI 2 to 36%).

At the 7 day follow-up, participants had a mean RMDQ change compared to baseline of 12.2 (95%CI 9.8 to 14.6) in the ibuprofen group, 9.5 (95%CI 7.1 to 12.0) in the diclofenac group, and 10.7 (95%CI 8.4 to 13.0) in the ibuprofen + diclofenac gel group. As shown in table 3, an analysis of the mean differences between the 3 groups showed no important differences. Slightly less than half of the study cohort reported taking medication for LBP at the 7 day follow-up (table 3). We report participant adherence to the invetsigational medications and use of off-protocol medications in the Appendix.

Table 3:

Seven-day outcomes

Variable	ibuprofen+ placebo gel	Diclofenac gel + placebo capsule	ibuprofen+ diclofenac gel		ibuprofen vs diclofenac gel	ibuprofen vs both	diclofenac vs both
	n=66	n=66	n=65		Mean difference, % (95% CI)
Mean RMDQ improvement	12.2 (9.8 to 14.6)a	9.5 (7.1 to 12.0)a	10.7 (8.4 to 13.0)a		2.7 (−0.7 to 6.1)b	1.6 (−1.8 to 4.9)b	−1.15 (−4.5 to 2.2)b
Worst LBP during previous 24h c
None	24 (40)	15 (24)	20 (33)
Mild	23 (38)	24 (39)	21 (34)		15% (0 to 31%)d	11% (−5 to 27%)d	4% (−13 to 21%)d
Moderate	9 (15)	16 (26)	13 (21)
Severe	4 (7)	7 (11)	7 (11)
Missing	6	4	4
Use of medication for LBP during previous 24 e
Yes	31 (52)	35 (56)	33 (54)		4% (−14 to 22%)	2% (−15 to 20%)	1% (−16 to 19%)
No	29 (48)	28 (44)	28 (46)
Missing	6	3	4

Abbreviations: RMDQ- Roland-Morris Disability Questionnaire, LBP- Low back pain

Data presented as No. (%) unless otherwise stated.

^aMean RMDQ improvement for patients in each group. The RMDQ improvement for each patient calculated as (baseline RMDQ score – RMDQ score at 7-day follow up)

^bThe between-group difference of the Mean RMDQ change at the 7-day follow up

^cWe asked patients “How would you describe your worst low back pain in the last 24 hours?”

^dNone/mild versus moderate/severe

^eWe asked patients “During the last 24 hours, have you taken any medication at all for low back pain?”

Adverse events were generally mild, uncommon, and occurred at similar frequencies among the 3 groups. In the ibuprofen group 3/60 (5%) participants reported adverse events. These include two patients with drowsiness and one with pruritis. In the diclofenac group, only 1/63 (2%) reported an adverse event (muscle cramps). In the combination arm, 4/64 (6%) reported adverse events. These include chills and diarrhea (in one participant), headaches, dizziness, and dyspepsia. The 95%CI for the rounded between group difference of 5% between diclofenac and the combination was −2, 11%.

LIMITATIONS

A number of limitations must be mentioned. The first limitation is we screened but did not include many patients because they did not meet our entry criteria. Therefore, the study participants represent only a subset of patients who present to the ED with acute non-traumatic, non-radicular LBP. These results are therefore most applicable to patients who meet our entry criteria.

A second limitation is that we conducted this study in one urban healthcare system serving a socio-economically depressed population. Because back pain outcomes may be associated with socioeconomic variables, our results can be generalized most appropriately to EDs that serve similar disadvantaged patient populations.

A third limitation is that we measured outcomes 7 days after the ED visit but only provided a 2-day supply of the medication (if in fact used every six hours). This discrepancy may have contributed to the lack of difference between the study groups at the 7-day follow-up.

A fourth limitation was our inability to mask the distinct odor of diclofenac gel. This could affect patient blinding if they could identify the investigational medication by its scent.

A fifth limitation was our use of the 4-point ordinal pain scale which, despite having face validity, has not been validated in the ED LBP patient population.

A sixth limitation was our selection of a 5-point change on the RMDQ scale as the minimum effect size required for clinical significance. There is debate regarding the optimal minimum change for clinical significance. (16). Furthermore, the RMDQ scale has not yet been validated in the ED setting.

DISCUSSION

In this randomized, double-blind comparative effectiveness study, topical diclofenac was probably not as efficacious as oral ibuprofen for patients with acute, musculoskeletal LBP who are discharged from an ED and it offered no additive benefit when combined with oral ibuprofen. Therefore, this study does not support the routine use of topical diclofenac among patients who can tolerate oral ibuprofen. However, the advantage of oral ibuprofen over topical diclofenac had dissipated one week after the ED visit. Most likely this dissipation reflects the natural history of acute LBP rather than late efficacy of topical diclofenac. Also of interest is the persistence of pain—more than ¼ of the entire cohort reported moderate or severe pain seven days after the ED visit and about 2/3rds reported some degree of pain.

It is important to understand the clinical relevance of the ibuprofen versus diclofenac comparison. Two days after discharge from the ED, the RMDQ score favored ibuprofen over diclofenac topical gel by 3.7 (95%CI:0.2, 7.2), a difference that was less than the minimum clinically important difference of five that we identified in the sample size calculation. However, biological phenomena are rarely explained by simple cut points. When one considers the distribution around this mean difference, it seems that the improvement in functional impairment experienced with diclofenac two days after the ED visit was less than that from ibuprofen. For some patients, this was clinically relevant and for others it was not. Furthermore, our analysis on pain intensity at two days reveals a statistically significant benefit favoring ibuprofen of 19%, corresponding to a number needed to treat of six.

Because of the lack of a pure placebo arm in this study, we do not know whether diclofenac is effective for acute musculoskeletal LBP. We only know that it is not as effective as oral ibuprofen and does not provide additive benefit when added to oral ibuprofen. It is clear that patients in the diclofenac + placebo arm improved during their participation in the study. However, we do not know whether treatment with diclofenac caused them to improve. Our study methodology does not provide any evidence as to whether topical NSAIDs should be used for patients with contra-indications to oral NSAIDs who present with low back pain.

Multiple studies have demonstrated the effectiveness of topical NSAIDS in the treatment of musculoskeletal disorders such as sprains, strains, contusions, and osteoarthritis-related pain (11). This has led the American College of Physicians to recommend topical NSAIDS as first-line treatment for these conditions (19). However, there is a paucity of evidence in using these medications for patients with acute low back pain. A previous open-label prospective cohort study suggested that topical diclofenac may provide relief for patients with acute back strain (13). A previous randomized controlled trial concluded that there was no benefit to adding topical diclofenac to topical capsaicin for acute back pain. However, this study included patients with neck pain in the primary analysis and reported outcomes of questionable clinical relevance (20). In an RCT measuring in-ED outcomes among patients with acute non-radiating low back pain, when compared to placebo, topical ketoprofen gel provided greater pain reduction when administered in addition to IV ketoprofen (14). However, this study did not report any outcomes beyond the first 30 minutes after medication administration.

Our data are consistent with current evidence that most patients with acute low back pain will improve with passage of time regardless of treatment (21, 22, 23). As frustrating as this is for patients and providers, management should focus not on treatment with medication but on addressing psychosocial factors, reassurance, and recommendations to avoid bedrest.

It is generally accepted that topical NSAIDs carry a lower risk of systemic adverse reactions compared to orally administered NSAIDs (22), due to significantly lower plasma concentrations following topical use (24, 25). Within the constraints of our small study population, adverse events were uncommon and occurred at similar rates among our three study groups, probably due to our broad exclusion of patients at higher risk of suffering adverse events from NSAIDs.

In conclusion, among patients with acute musculoskeletal low back pain discharged from an ED, topical diclofenac was probably less efficacious than oral ibuprofen and demonstrated no additive benefit when co-administered with oral ibuprofen.