Comparative effectiveness of amitriptyline vs duloxetine in the treatment of chronic low back pain: An observational study

Abstract

BACKGROUND

Chronic nonspecific low back pain is defined as pain that persists for greater than 12 weeks and mainly occurs in the lower back with no evidence of associated underlying serious conditions [like malignancy, inflammation (like ankylosing spondylitis) or infection, vertebral fracture, etc.].

AIM

To compare the efficacy and safety of amitriptyline with duloxetine in treating chronic low back pain (CLBP).

METHODS

The present study was a two-arm observational study conducted over 18 months in a tertiary rehabilitation setting. A total of 254 patients were included in the study.

RESULTS

The mean age was significantly higher in the amitriptyline group (34.78 ± 8.22 years) compared with the duloxetine group (29.98 ± 7.28 years, P < 0.0001). Baseline visual analog scale (VAS) scores were also significantly different between groups (amitriptyline: 7.92 ± 0.56; duloxetine: 8.46 ± 0.79; P < 0.0001). Within-group analysis showed a significant reduction in VAS scores over time in both groups (P < 0.001). At 12 weeks the duloxetine group showed significantly lower VAS scores (0.92 ± 0.78) compared with the amitriptyline group (1.87 ± 1.71; P < 0.0001). Analysis of variance, adjusting for age and baseline VAS, confirmed a significant group effect on pain reduction at 12 weeks (P < 0.001), favoring duloxetine. Side effects were generally mild. The most common in the amitriptyline group were dry mouth (17.3%) and drowsiness (7.9%) while in the duloxetine group, dry mouth (15.7%) and constipation (2.4%) were most reported.

CONCLUSION

Amitriptyline and duloxetine effectively treat CLBP; however, considering side effects and more sustained pain relief, duloxetine appears to be the better option. Nonetheless, treatment choice should consider individual patient profiles.

Key Words: Amitriptyline; Duloxetine; Chronic low back pain; Neuropathic pain; Pain

Core Tip: This study compared the efficacy and safety of amitriptyline and duloxetine in treating chronic low back pain. Both medications were effective in reducing pain as evidenced by statistically significant improvements in the visual analog scale at 1 week, 4 weeks, and 12 weeks. However, duloxetine demonstrated a more favorable side effect profile and better long-term sustained effects with fewer anticholinergic and sedative effects compared with amitriptyline. These findings suggest that while both drugs are viable options for chronic low back pain management, duloxetine may be preferred for patients seeking a treatment with fewer side effects. The choice of medication should be individualized based on patient-specific factors and health conditions.

INTRODUCTION

Chronic low back pain (CLBP) is a disabling condition and one of the most common musculoskeletal disorders in developed countries. It imposes a significant economic burden due to its detrimental effects on adults of working age. Chronic nonspecific low back pain (LBP) is defined as pain lasting longer than 12 weeks and primarily occurring in the lower back with no evidence of serious underlying conditions (such as malignancy, ankylosing spondylitis, infection, or vertebral fractures). The etiology of CLBP appears to be multifactorial. While it is not fully understood, it has a profound adverse impact on patient quality of life. Diagnosing CLBP is complex, and treatment typically involves a multidisciplinary approach, which may be pharmacological or non-pharmacological[1].

Compared with amitriptyline, which was launched in 1961, duloxetine is a more recent drug released in 2004. Duloxetine has the advantage of a safer profile with fewer anticholinergic side effects such as dry mouth, tachycardia, constipation, and weight gain, among others. Systematic reviews indicate that both amitriptyline and duloxetine are effective in treating CLBP[2].

LBP is the largest contributor to disability worldwide[3]. Although a variety of treatments are available for LBP, the efficacy of these therapies is often limited. Antidepressants are commonly prescribed to manage LBP in clinical practice. Typically, higher doses of antidepressants are used to treat depression while lower doses are prescribed for chronic pain with the analgesic effects occurring independently of depression. The use of antidepressants has been rapidly increasing with a 6.8% rise in prescriptions (3.9 million) in the United Kingdom over 12 months, and 29% of these prescriptions being off-label (for unapproved indications)[4]. This increase persists despite the lack of evidence from systematic reviews and conflicting recommendations in clinical guidelines.

A review of both national and international guidelines has highlighted substantial variability in recommendations for antidepressant treatment of LBP. Seven out of fourteen guidelines recommend their use, but none specify whether they should be prescribed in high or low doses[5]. Two treatment guidelines published between 2016 and 2017 provide further conflicting recommendations. One guideline suggests that while tricyclic antidepressants (TCAs) and selective serotonin reuptake inhibitors are not recommended, duloxetine hydrochloride should be considered a second-line therapy. The other guideline does not recommend the use of any antidepressant class for chronic LBP[6].

Numerous systematic reviews, including the Cochrane systematic review, have concluded that there is no clear evidence that antidepressants are more effective than placebo for LBP[7]. These reviews highlight the need for high-quality trials and point out limitations in previous studies, including insufficient blinding, small sample sizes, and short treatment and follow-up periods (≤ 3 months). Furthermore, no studies have specifically examined the effectiveness of low-dose TCAs, a common method of prescribing for LBP. Amitriptyline hydrochloride, a TCA, is widely used in low doses to treat pain, particularly nonspecific LBP, independent of depression[8]. However, there is no strong evidence supporting its widespread use.

Given these uncertainties, we began our study to compare the effectiveness of both drugs in treating CLBP. This study aimed at to compare the efficacy and safety of amitriptyline with duloxetine in CLBP by evaluating pain intensity (primary objective by visual analogue scale), pain severity and its impact on daily life [secondary objective by brief pain inventory (BPI–short form)].

MATERIALS AND METHODS

This study was initiated after receiving the approval from the Institutional Ethical Committee of R. G. Kar Medical College and Hospital (No. RKC/687), Kolkata, West Bengal, India. Amitriptyline and duloxetine were given free of cost to the patient from the Out Patient Department Dispensary of R. G. Kar Medical College and Hospital. History, clinical examinations with relevant information, and measurements were noted in pretested, predesigned study proforma.

All patients and their caregivers were informed of their physical status, available treatment, outcomes, and expected adverse effects in a language that was understandable to them. Informed consent was obtained from each participant before inclusion in the study. This was a two-arm observational study. This study was conducted in the Department of Physical Medicine and Rehabilitation. The study was conducted for 18 months. All patients who were clinically diagnosed, met the inclusion criteria, and did not exhibit any of the exclusion criteria were included in the study population.

Inclusion criteria

Patients were enrolled based on the following inclusion criteria: Age ≥ 18 years and ≤ 75 years; diagnosis of CLBP based on a combination of clinical signs and symptoms; and patients with failed conservative treatment other than antidepressants for the last 3 months.

Exclusion criteria

Patients were excluded based on the following criteria: Presence of depression (either major or minor) or anxiety (moderate to severe) disorder according to the criteria of the Diagnostic and Statistical Manual, 5^th edition; history of trauma or fracture in the lower back; LBP with neurological deficits; history of ischemic heart disease or cardiac arrhythmia; glaucoma; urological disturbances; chronic constipation; indications for epidural steroid injection; and pregnancy.

Sample size

Based on the result of the previous study “Efficacy of low dose amitriptyline for chronic low back pain”, the sample size of each matched comparable group of this study was ascertained by this formula[9]: N = [(Z_1-α + Z_1-β)² × 2 × S²]/δ²; where α is the type I error rate and β is the type II error rate, N is the sample size of each group, δ = 1, difference of means of the two groups is Z_1-α = 1.96 [Consider 95% confidence interval (CI), Z_1-β = 0.84 (at 80% power of study)], and S = 2.7 [standard deviation (SD)]. Therefore, N = [(1.96 + 0.84)² × 2 × S²]/1² = 114.30 = 114 (rounded). Estimating a 10% drop out rate, the total sample size was 2 × 127 = 254.

Study variable and parameters to be measured

The outcome measurements depended on different variables of the study. Apart from the sociodemographic and other clinical variables, the specific variables of the study were pain intensity and quality of life.

Independent variables: Age, gender, education, occupation, socioeconomic status, addiction, and duration of pain.

Dependent variables: Pain intensity [visual analog scale (VAS)], the severity of pain, and its impact on daily life (BPI-short form).

Procedure of study/data collection

Selection and identification of cases: After getting the institutional clearance, all the patients who satisfied the inclusion criteria clinically and did not have any features of exclusion criteria were identified. Written informed consent was obtained from all participants in their own language after explaining the purpose of the study and confidentiality of their information. Patient’s particulars, clinical history, and examination findings were recorded on a preformed, pretested proforma. A double-blinding method was adopted in selecting the cases, and sealed envelope was used to conceal the allocations to groups. Persons involved in selection, group allocation, and measuring outcomes were all different and were blinded.

Study design: An observational study was carried out using the hospital database (rehabilitation settings) for CLBP. Patients with CLBP who received amitriptyline (25 mg/day) and duloxetine (60 mg/day) were screened. Cases that continued the same dose of medications were included. Adherence was ensured by maintaining medication diary by participants.

All patients were asked for the level of pain using the VAS (0-10) at the baseline (visit 0), at 1 week (visit 1), 4 weeks (visit 2), and 12 weeks (visit 3).

Statistical analysis

For statistical analysis data were entered into a Microsoft excel spreadsheet and then analyzed by SPSS (version 27.0; SPSS Inc., Chicago, IL, United States) and GraphPad Prism version 5. Data had been summarized as mean and SD for numerical variables and count and percentages for categorical variables. Two-sample t-tests for a difference in mean involved independent samples or unpaired samples. Paired t-tests were a form of blocking and had greater power than unpaired tests. The χ² test was any statistical hypothesis test wherein the sampling distribution of the test statistic was a χ² distribution when the null hypothesis was true. Without other qualification χ² test is short for Pearson’s χ² test. Unpaired proportions were compared by χ² test or Fischer’s exact test as appropriate.

Explicit expressions that can be used to carry out various t-tests are given below. In each case the formula for a test statistic that either exactly followed or closely approximated a t-distribution under the null hypothesis was given. The appropriate degrees of freedom were given in each case. Each of these statistics can be used to carry out either a one-tailed test or a two-tailed test.

Once a t value was determined, a P value was found using a table of values from the Student’s t-distribution. If the calculated P value was below the threshold chosen for statistical significance (usually the 0.10, the 0.05, or 0.01 level), then the null hypothesis was rejected in favor of the alternative hypothesis.

Analysis of variance (ANOVA) was also conducted to account for baseline differences in key variables, particularly age and baseline VAS scores, which could confound the treatment effects on pain reduction. By adjusting for these covariates, ANOVA allowed us to more accurately assess the independent effect of the treatment group (amitriptyline vs duloxetine) on pain outcomes at 12 weeks, ensuring that any observed differences were not merely due to preexisting group disparities. A P value ≤ 0.05 was statistically significant.

RESULTS

Demographic and clinical characteristics

A total of 254 participants were enrolled with 127 patients in the amitriptyline group and 127 patients in the duloxetine group.

Age: In the amitriptyline group the mean age (mean ± SD) of patients was 34.7795 ± 8.2201. In the duloxetine group the mean age (mean ± SD) of patients was 29.9843 ± 7.2801. The distribution of the mean age was statistically significant (P < 0.0001).

Age groups: In the amitriptyline group 5 (3.9%) patients were ≤ 20 years of age, 39 (30.7%) patients were 21-30 years of age, 63 (49.6%) patients were 31-40 years of age, and 20 (15.7%) patients were ≥ 51 years of age. In the duloxetine group 5 (3.9%) patients were ≤ 20 years of age, 58 (45.7%) patients were 21-30 years of age, 55 (43.3%) patients were 31-40 years of age, and 9 (7.1%) patients were 41-50 years of age. The association of age group was statistically significant (P < 0.0001).

Sex: In the amitriptyline group 34 (26.8%) patients were female, and 93 (73.2%) patients were male. In the duloxetine group 49 (38.6%) patients were female, and 78 (61.4%) patients were male. The association of sex with treatment was statistically significant (P = 0.04478).

Occupations: In the amitriptyline group 15 (11.8%) patients were carpenters, 5 (3.9%) patients were cricketers, 48 (37.8%) patients were farmers, 24 (18.9%) patients were housewives, 10 (7.9%) patients were maid servants, 15 (11.8%) patients were masons, and 10 (7.9%) patients were painters. In the duloxetine group 14 (11.0%) patients were carpenters, 5 (3.9%) patients were cricketers, 31 (24.4%) patients were farmers, 29 (22.8%) patients were housewives, 20 (15.7%) patients were maid servants, 19 (15.0%) patients were masons, and 9 (7.1%) patients were painters. The association of occupation with treatment was not statistically significant (P = 0.2366).

Duration of CLBP: In the amitriptyline group the mean duration (mean ± SD) of CLBP was 14.1339 ± 1.2495 weeks. In the duloxetine group, the mean duration (mean ± SD) of CLBP was 14.1811 ± 1.1576 weeks. The distribution of the mean duration with the group was not statistically significant (P = 0.7549).

Side effects: In the amitriptyline group of 127 patients, 22 patients (17.32%) experienced dry mouth, 10 patients (7.87%) reported drowsiness, 2 patients (1.57%) had dizziness, and 2 patients (1.57%) experienced constipation.

In the duloxetine group consisting of 127 patients, 20 patients (15.75%) reported dry mouth, 5 patients (3.94%) experienced drowsiness, 1 patient (0.79%) had dizziness, 2 patients (1.57%) experienced nausea, and 3 patients (2.36%) reported constipation. Since side effects were mild, there were no drop-outs.

Statistical adjustments

Age and VAS baseline: ANOVA was conducted to examine the effect of groups on VAS baseline, controlling for age. A preliminary analysis evaluating the homogeneity-of-slopes assumption indicated that the relationship between the covariate and the dependent variable did not differ significantly as a function of the independent variable. A weak negative correlation was found between the covariate and dependent variable, r = -0.13. After adjustment for age there was a statistically significant difference in VAS baseline between groups, F (1, 251) = 38.00, P = 0.001, partial η² = 0.131. The group factor accounted for 13.1% of the variance in VAS baseline, holding constant age. The covariate, age, was significantly related to VAS baseline, F (1, 251) = 0.06, P = 0.810, partial η² = 0.000. Adjusted mean (M) scores were: Amitriptyline (M = 7.92) and duloxetine (M = 8.48). Follow-up analyses were conducted to examine the pairwise differences among adjusted means. The adjusted mean for duloxetine (M = 0.55, 95%CI: 0.38-0.72) was significantly different from amitriptyline (P < 0.001).

Age and VAS at 12 weeks: ANOVA was conducted to examine the effect of the group on VAS at 12 weeks, controlling for age. A preliminary analysis evaluating the homogeneity-of-slopes assumption indicated that the relationship between the covariate and the dependent variable did not differ significantly as a function of the independent variable. A weak positive correlation was found between the covariate and dependent variable, r = 0.09. After adjustment for age, there was a statistically significant difference in VAS at 12 weeks between groups, F (1, 251) = 29.75, P = 0.001, partial η² = 0.106. The group factor accounted for 10.6% of the variance in VAS at 12 weeks, holding constant age. The covariate, age, was significantly related to VAS at 12 weeks, F (1, 251) = 0.01, P = 0.910, partial η² = 0.000. Adjusted mean scores were: Amitriptyline (M = 1.88) and duloxetine (M = 0.92). Follow-up analyses were conducted to examine the pairwise differences among adjusted means. The adjusted mean for duloxetine (M = -0.96, 95%CI: -1.29 to -0.63) was significantly different from amitriptyline (P < 0.001).

VAS baseline and VAS at 12 weeks: ANOVA was conducted to examine the effect of the groups on VAS at 12 weeks, controlling for VAS baseline. A preliminary analysis evaluating the homogeneity-of-slopes assumption indicated that the relationship between the covariate and the dependent variable did not differ significantly as a function of the independent variable. A weak negative correlation was found between the covariate and dependent variable, r = -0.19. After adjustment for VAS baseline, there was a statistically significant difference in VAS at 12 weeks between groups, F (1, 251) = 23.05, P = 0.001, partial η² = 0.084. The group factor accounted for 8.4% of the variance in VAS 12 weeks after treatment, holding constant VAS baseline. The covariate, VAS baseline, was significantly related to VAS at 12 weeks, F (1, 251) = 1.46, P = 0.227, partial η² = 0.000. Adjusted mean scores were: Amitriptyline (M = 1.83) and duloxetine (M = 0.96). Follow-up analyses were conducted to examine the pairwise differences among adjusted means. The adjusted mean for duloxetine (M = -0.87, 95%CI: -1.20 to -0.54) was significantly different from amitriptyline (P < 0.001).

Outcome measures

Baseline VAS: In the amitriptyline group the mean VAS at baseline (mean ± SD) of patients was 7.9213 ± 0.5579. In the duloxetine group the mean VAS at baseline (mean ± SD) of patients was 8.4646 ± 0.7948. The distribution of mean VAS at baseline with the group was statistically significant (P < 0.0001).

VAS at 1 week: In the amitriptyline group the mean VAS 1 week after treatment (mean ± SD) of patients was 0.9145 ± 0.6375. In the duloxetine group the mean VAS 1 week after treatment (mean ± SD) of patients was 4.0394 ± 1.2306. The distribution of mean VAS 1 week after treatment with the group was statistically significant (P < 0.0001).

VAS at 4 weeks: In the amitriptyline group the mean VAS 4 weeks after treatment (mean ± SD) of patients was 1.0000 ± 0.9913. In the duloxetine group the mean VAS 4 weeks after treatment (mean ± SD) of patients was 1.5984 ± 0.9450. The distribution of mean VAS 4 weeks after treatment with the group was statistically significant (P < 0.0001).

VAS at 12 weeks: In the amitriptyline group the mean VAS 12 weeks after treatment (mean ± SD) of patients was 1.8740 ± 1.7182. In the duloxetine group the mean VAS 12 weeks after treatment (mean ± SD) of patients was 0.9213 ± 0.7828. The distribution of mean VAS 12 weeks after treatment with the group was statistically significant (P < 0.0001). Changes from baseline to 12 weeks was also statistically significant in both the groups (P < 0.0001).

Baseline BPI: In the amitriptyline group the mean BPI at baseline (mean ± SD) of patients was 8.5118 ± 1.1606. In the duloxetine group the mean BPI at baseline (mean ± SD) of patients was 8.2913 ± 1.0624. The distribution of the mean BPI at baseline with the group was not statistically significant (P = 0.1156).

BPI at 1 week: In the amitriptyline group the mean BPI 1 week after treatment (mean ± SD) of patients was 6.4016 ± 1.2036. In the duloxetine group the mean BPI 1 week after treatment (mean ± SD) of patients was 6.3701 ± 1.0374. The distribution of the mean BPI 1 week after treatment with the group was not statistically significant (P = 0.8234).

BPI at 4 weeks: In the amitriptyline group the mean BPI 4 weeks after treatment (mean ± SD) of patients was 4.4409 ± 0.7309. In the duloxetine group the mean BPI 4 weeks after treatment (mean ± SD) of patients was 4.4724 ± 0.7328. The distribution of the mean BPI 4 weeks after treatment with the group was not statistically significant (P = 0.7319).

BPI at 12 weeks: In the amitriptyline group the mean BPI 12 weeks after treatment (mean ± SD) of patients was 0.1803 ± 0.7109. In the duloxetine group the mean BPI 12 weeks after treatment (mean ± SD) of patients was 2.0157 ± 0.6899. The distribution of the mean BPI 12 weeks after treatment with the group was not statistically significant (P = 0.1803) (Table 1).

Table 1 Changes of outcome measures at different time points.

Outcome measures	Groups	Mean	SD	P value
VAS at baseline	Amitriptyline	7.9213	0.5579	< 0.0001
	Duloxetine	8.4646	0.7948
VAS at 1 week	Amitriptyline	0.9145	0.6375	< 0.0001
	Duloxetine	4.0394	1.2306
VAS at 4 weeks	Amitriptyline	1.0000	0.9913	< 0.0001
	Duloxetine	1.5984	0.9450
VAS at 12 weeks	Amitriptyline	1.8740	1.7182	< 0.0001
	Duloxetine	0.9213	0.7828
BPI at baseline	Amitriptyline	8.5118	1.1606	0.1156
	Duloxetine	8.2913	1.0624
BPI at 1 week	Amitriptyline	6.4016	1.2036	0.8234
	Duloxetine	6.3701	1.0374
BPI at 4 weeks	Amitriptyline	4.4409	0.7309	0.7319
	Duloxetine	4.4724	0.7328
BPI at 12 weeks	Amitriptyline	0.1803	0.7109	0.1803
	Duloxetine	2.0157	0.6899

VAS: Visual analog scale; BPI: Brief pain inventory; SD: Standard deviation.

DISCUSSION

This was the first study that provided a comparative analysis of amitriptyline and duloxetine in the treatment of CLBP, focusing on pain intensity reduction over a 12-week period. The results demonstrated that both drugs were effective but with distinct patterns in pain reduction and response timelines.

Importantly, the duloxetine group reported significantly higher baseline pain scores (VAS) even after controlling for age. While this initially might have biased the results, follow-up ANOVA models adjusting for both age and baseline VAS confirmed that duloxetine still produced greater pain reduction at 12 weeks than amitriptyline. These results underscore the robustness of the long-term efficacy of duloxetine, independent of initial pain severity or demographic differences. A previous study also showed effectiveness with the same dose of duloxetine[10]. A previous study with amitriptyline showed improvement in back pain but did not show statistically significant improvements[9].

Although amitriptyline showed a more rapid early response (e.g., significant pain reduction at 1 and 4 weeks), duloxetine was associated with significantly lower pain scores at the 12-week mark. The adjusted mean difference of nearly 1 point on the VAS scale favored duloxetine with moderate effect sizes (η²: 0.08–0.13), indicating a clinically meaningful advantage for duloxetine in the longer term.

The role of age and baseline VAS as covariates was statistically non-significant in all models, suggesting that group differences in treatment outcome were driven primarily by the treatment itself, not by confounding demographic variables. This strengthens the internal validity of the results.

From a tolerability standpoint both drugs had relatively mild and infrequent side effects. Amitriptyline was associated with more sedative effects, such as drowsiness and dry mouth, while duloxetine showed fewer early side effects and might thus offer better long-term adherence potential for some patient populations. In CLBP both drugs likely exert their benefits by enhancing descending pain inhibition through serotonin and norepinephrine modulation. Duloxetine has a more selective mechanism with fewer off-target effects while amitriptyline has a broader pharmacological profile that may provide additional sedative benefit but also increases side-effect burden. Previous systematic reviews also showed similar side-effect profiles[10,11].

However, this study had limitations. The absence of a placebo or control group limits causal inference as the placebo effect, known to be substantial in pain research, cannot be ruled out. Additionally, although the ANOVA approach helped control for certain confounders, residual confounding and the influence of unmeasured variables (e.g., psychosocial factors, comorbidities) may still impact the findings.

CONCLUSION

Both amitriptyline and duloxetine are effective in reducing pain in CLBP, but duloxetine provided more sustained relief at 12 weeks even after controlling for age and initial pain severity. These findings support duloxetine as a potentially more durable treatment option in CLBP, particularly when long-term management is the goal. Future randomized, placebo-controlled trials with larger and more diverse populations are recommended to confirm these findings and better delineate patient subgroups that may benefit most from each treatment.