A comprehensive study on drug-related Raynaud’s phenomenon based on the FDA adverse event reporting system

Abstract

Raynaud’s phenomenon (RP) is often an overlooked adverse event, mainly secondary RP, where drug induction or exacerbation is a controllable and preventable factor. This study aimed to systematically evaluate the association between drugs and RP using the FDA adverse event reporting system (FAERS) database. Utilizing disproportionality analysis, we quantified the risk of RP-associated drugs based on large-scale FAERS case data. Drugs were categorized by ATC classification, and descriptive analyses were conducted to assess population characteristics and reporting patterns. Secondary analyses and Weibull shape parameter testing were performed to verify the reliability of the results and the temporal patterns of adverse event onset. A total of 562 drugs and 4303 cases were linked to RP. Significant signals were identified across 8 ATC categories, predominantly nervous system agents (21 drugs, n = 632), cardiovascular drugs (20 drugs, n = 403), and immunomodulators (14 drugs, n = 185). Pharmacologically, CGRP inhibitors, triptans, non-selective β-blockers, interferons, and antineoplastic/immunomodulatory agents demonstrated high mechanistic relevance to RP. Indications such as rheumatoid arthritis (256 cases, 5.95%), narcolepsy (149 cases, 3.46%), and attention-deficit hyperactivity disorder (117 cases, 2.72%) were frequently associated with RP. Cases were concentrated in individuals aged 25–59, with a female predominance (ratio 3.13:1). This study systematically reports drug-RP associations. Clinicians should consider RP risk when prescribing CGRP inhibitors, triptans, non-selective β-blockers, interferons, or antineoplastic/immunomodulatory agents. Middle-aged and elderly females, particularly those with rheumatoid arthritis, warrant heightened vigilance for drug-induced RP.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-17182-z.

Introduction

Raynaud’s phenomenon (RP) is an exaggerated vascular response to cold or emotional stress, clinically characterized by a classic “triphasic color change” in the extremities under such triggers as pallor (ischemia), cyanosis (hypoxia), and erythema (reperfusion)^1,2. This process is often accompanied by needling pain, numbness, and reduced temperature sensation, with severe cases progressing to digital ulcers or even irreversible necrosis³, significantly impairing daily function and quality of life. RP is classified into primary and secondary forms⁴. Primary RP typically presents with milder symptoms unaccompanied by tissue damage. It predominantly affects women under the age of 40. Some cases may resolve spontaneously. Nailfold capillaroscopy findings are normal. In contrast, secondary RP is induced by an underlying condition or causative factor. Symptoms are more severe and may manifest as asymmetric attacks, pain, and tissue damage (e.g., ulcers or gangrene). Onset typically occurs after the age of 40. This form is frequently associated with abnormal capillaroscopic patterns, positive clinical findings, or laboratory indicators suggestive of the underlying pathology⁵. The global prevalence of primary RP varies geographically, ranging from 2 to 12% in temperate regions to as high as 17% in colder Nordic areas⁶. A large UK cohort study reported primary Raynaud’s phenomenon (RP) in 91.7% of cases versus 8.3% secondary RP, with higher female incidence. Mean diagnosis ages were 44.5 years (women) and 50.2 years (men) for primary RP, rising to 55.6 (women) and 57.8 years (men) for secondary RP. The 2021 adult primary RP incidence was 32/100,000 person-years, with winter rates 3–6 times summer rates. Prevalence was an estimated 0.5% in adults and 0.75% in elderly populations⁷. Although primary RP accounts for the majority of cases (80–90%)⁸, secondary RP may arise from autoimmune diseases (e.g., systemic lupus erythematosus, vasculitis) or systemic disorders, with studies indicating that over 96% of systemic sclerosis patients develop concurrent RP⁹. Drugs represent a critical and modifiable factor in inducing or exacerbating secondary RP, primarily via mechanisms such as vascular tone imbalance, endothelial dysfunction, and abnormal neural regulation. Over 40 drugs have been reported in association with RP¹⁰; however, evidence predominantly stems from small-sample retrospective studies or isolated case reports, lacking systematic evaluation or quantitative analysis. Furthermore, existing literature inadequately explores drug-induced RP’s mechanisms, risk factors, and epidemiological features, particularly studies leveraging large-scale real-world data.

The FDA adverse event reporting system (FAERS) is an open global database containing millions of adverse drug event reports, providing a valuable data resource for drug safety monitoring and research. FAERS data primarily consists of post-marketing spontaneous reports, supplemented by a smaller proportion of reports submitted during clinical trials by sponsors or investigators. Reports are submitted through multiple channels, including voluntary submissions from healthcare professionals and consumers, as well as individual case safety reports (ICSRs) submitted by marketing authorization holders (MAHs) in compliance with the ICH E2B(M) standard via the FDA’s electronic submissions gateway (ESG) or safety reporting portal (SRP). Maintained by the U.S. Food and Drug Administration (FDA), FAERS accepts adverse event reports from worldwide sources. Its primary purpose is to address the inherent limitations of pre-marketing clinical studies. Particularly, reports submitted by healthcare professionals are generally considered to possess higher reliability due to their specialized clinical expertise; this perception stems from enhanced accuracy in adverse event documentation and causality assessment capabilities.

Disproportionality analysis, a widely used data mining method in pharmacovigilance, identifies potential associations between drugs and adverse events¹¹. Compared to traditional clinical trials or observational studies, FAERS data combined with disproportionality analysis enables rapid and efficient detection of drug-related adverse event signals, offering critical insights for drug safety evaluation. This study aims to systematically evaluate drug signals associated with RP using disproportionality analysis based on the FAERS database. Descriptive analyses will characterize population features, while secondary analyses will strengthen the evidence base. The findings are intended to enhance clinical and research understanding of drug-related RP and provide scientific support for drug safety monitoring and risk management.

Results

Basic characteristics of Raynaud’s phenomenon cases and population

From the first quarter of 2004 to the second quarter of 2024, the FAERS recorded a total of 21,558,936 adverse event case reports. After deduplication and cleaning, 17,956,653 valid cases remained, among which 4237 involved RP-related drugs, accounting for 0.0239% of all adverse event reports associated with PS medications. Following deduplication, 562 unique drugs were identified, associated with 4303 cases. As illustrated in Fig. 1A, given that the 2024 data represent only two quarters and cannot reliably reflect annual trends, linear regression analysis was performed exclusively on reported RP cases from 2004 to 2023. The results demonstrate a statistically significant upward trend, with an average annual increase of 16.3 cases (p < 0.001). The coefficient of determination (R² = 0.90) indicates that temporal changes explain 90% of the variability in RP case reports. The highest number of reports was recorded in 2023 with 456 cases (10.62%), while the lowest number occurred in 2004 with 76 cases (1.77%). Figure 1B displays the top 10 countries with the highest number of reported RP cases. The United States (52.22%), the United Kingdom (10.43%), and Canada (8.17%) ranked as the top three source countries for reported cases. However, this outcome does not imply a lower incidence in other countries, but it is closely tied to reporting rates and the maturity of regulatory systems. Figure 1C categorizes reporters of RP cases. Healthcare professionals (HCPs), including physicians, pharmacists, and other allied health professionals, reported 2490 cases, while non-healthcare reporters (e.g., consumers, lawyers) reported 1552 cases, giving a ratio of 1.60:1. Among HCPs, physicians accounted for the majority of reports; among non-healthcare reporters, consumers were the primary contributors.

Fig. 1.

(A) Annual reporting trends of RP; (B) National reporting distribution of RP; (C) Reporter demographics for RP (healthcare professionals vs. non-healthcare reporters); (D) Age and gender distribution of RP cases; (E) Outcomes; CA congenital anomaly, DE death, DS disability, HO hospitalization or prolongation of hospitalization, LT life-threatening, OT other serious, RI required intervention to prevent permanent impairment/damage, F female, M male.

Analysis of age and gender distribution among RP cases (Fig. 1D) revealed that the 25–59-year age group represented the highest proportion of RP cases (1520 cases; 54.85%), followed by the 60–79-year group (893 cases; 32.23%). The female-to-male ratio was 3.13:1, indicating a significantly higher prevalence among females. Figure 1E illustrates the outcomes of RP across different age groups. “Other outcomes” constituted the largest proportion in all age groups, followed by prolonged hospitalization and disability. This trend was particularly pronounced in the 18–64-year-old population. Analysis of patient indications (Fig. 2) identified the top five conditions associated with RP: rheumatoid arthritis (256 cases, 5.95%), narcolepsy (149 cases, 3.46%), attention-deficit/hyperactivity disorder (ADHD) (117 cases, 2.72%), multiple sclerosis (116 cases, 2.70%), and pulmonary arterial hypertension (111 cases, 2.58%).

Fig. 2.

Proportion of indications (number of cases, percentage of total cases).

Drug characteristics associated with Raynaud’s phenomenon and secondary analysis

Among the 562 drugs linked to RP, the top 50 medications by frequency are summarized in Table 1. Notable examples include: Oxybate sodium (172 cases, 0.26% of all adverse reactions for this drug), Adalimumab (136 cases, 0.02%), Etanercept (134 cases, 0.03%), Lisdexamfetamine (100 cases, 0.59%), Interferon beta-1a (84 cases, 0.06%). Based on the proportion of RP cases relative to all reported adverse events for each drug, lisdexamfetamine demonstrated a non-maximal frequency of 100 cases among the top 50 medications but exhibited a comparatively higher RP-associated proportion of 0.59%. Among the top 50 medications analyzed by ATC classification, immunomodulating agents comprised 21 drugs accounting for 972 cases (22.94%); cardiovascular agents included 10 drugs with 387 cases (9.13%); and nervous system agents involved 9 drugs representing 586 cases (13.83%). This distribution highlights a significant frequency-based association between immunomodulating agents and the occurrence of RP.

Table 1.

Number of drug cases involving Raynaud’s phenomenon and their proportion in all adverse reactions.

Information of the top 50 drugs associated with RP frequency	Cases	Percentage of RP in all adverse reactions of the drug (%)	First-level classification of ATC
Aripiprazole	22	0.03	Nervous
Atomoxetine	44	0.25	Nervous
Erenumab	55	0.12	Nervous
Galcanezumab	42	0.18	Nervous
Lisdexamfetamine	100	0.59	Nervous
Methylphenidate	70	0.20	Nervous
Oxybate sodium	172	0.26	Nervous
Pregabalin	26	0.02	Nervous
Sumatriptan	35	0.16	Nervous
Amfetamine; Dexamfetamine	20	0.14	Nervous
Alendronic acid	80	0.24	Musculo-skeletal
Denosumab	31	0.02	Musculo-skeletal
Rofecoxib	35	0.10	Musculo-skeletal
Adalimumab	136	0.02	Immunomodulating
Belimumab	23	0.11	Immunomodulating
Capecitabine	20	0.04	Immunomodulating
Carboplatin	26	0.06	Immunomodulating
Certolizumab	27	0.04	Immunomodulating
Etanercept	134	0.03	Immunomodulating
Gemcitabine	42	0.17	Immunomodulating
Infliximab	68	0.04	Immunomodulating
Interferon beta-1a	84	0.06	Immunomodulating
Interferon beta-1b	21	0.09	Immunomodulating
Lenalidomide	28	0.01	Immunomodulating
Methotrexate	84	0.07	Immunomodulating
Mycophenolic acid	34	0.06	Immunomodulating
Natalizumab	24	0.02	Immunomodulating
Ocrelizumab	22	0.05	Immunomodulating
Paclitaxel	24	0.07	Immunomodulating
Peginterferon alfa-2a	21	0.07	Immunomodulating
Rituximab	53	0.06	Immunomodulating
Teriflunomide	25	0.06	Immunomodulating
Tocilizumab	23	0.04	Immunomodulating
Tofacitinib	53	0.04	Immunomodulating
Ciprofloxacin	29	0.08	Antiinfectives
Immunoglobulins, unspecified	20	0.03	Antiinfectives
Sildenafil	24	0.05	Genitourinary
Intrauterine contraceptive device; Levonorgestrel	20	0.02	Genitourinary
Ambrisentan	38	0.04	Cardiovascular
Amlodipine	31	0.07	Cardiovascular
Atorvastatin	50	0.06	Cardiovascular
Bisoprolol	56	0.52	Cardiovascular
Bosentan	43	0.11	Cardiovascular
Lisinopril	32	0.12	Cardiovascular
Macitentan	38	0.10	Cardiovascular
Metoprolol	21	0.08	Cardiovascular
Ramipril	56	0.33	Cardiovascular
Simvastatin	22	0.08	Cardiovascular
Treprostinil	46	0.06	Blood
Esomeprazole	23	0.03	Alimentary and metabolism

Relying solely on drug frequency to evaluate associations between medications and RP risks introducing multiple biases. Using mainstream disproportionality analysis, 562 drugs meeting the minimum threshold of ≥ 3 RP cases were screened using four algorithms: ROR, PRR, BCPNN, MGPS. Only 68 drugs were identified as positive signal drugs with strong associations to RP. A forest plot (Fig. 3) visualizes the top 50 drugs by frequency, along with their ROR values and 95% confidence intervals (CIs). Among the top five drugs, three are nervous system agents: including oxybate sodium (ROR 11.36, 95% CI 9.75–13.23), lisdexamfetamine (ROR 25.19, 95% CI 20.65–30.74), alendronic acid (ROR 10.14, 95% CI 8.12–12.65), methylphenidate (ROR 8.28, 95% CI 6.54–10.49), bisoprolol (ROR 22.13, 95% CI 16.99–28.83). Labetalol exhibited the highest risk signal (ROR 55.95, 95% CI 33.59–93.2). Table 2 details the signal values from all four methods and the ATC classifications for the 68 identified drugs. Stratification by ATC classification revealed significant pharmacovigilance signals across multiple drug categories. For nervous system agents (21 drugs; 632 RP cases, 43.86%), key signals included triptans—sumatriptan (ROR 6.89, 95% CI4.94–9.61), rizatriptan (18.82, 9.77–36.24), eletriptan (7.77, 3.49–17.31)—and calcitonin gene-related peptide (CGRP) inhibitors: erenumab (5.22, 4.00–6.81), fremanezumab (8.34, 5.02–13.85), galcanezumab (7.78, 5.74–10.55), atogepant (5.11, 1.91–13.62), rimegepant (4.68, 2.34–9.37). Cardiovascular agents (20 drugs; 403 cases, 27.97%) demonstrated elevated risks with β-blockers—propranolol (8.04, 5.18–12.49), nadolol (18.91, 6.08–58.79), carteolol (69.47, 25.85–186.69), labetalol (55.95, 33.59–93.20); ACE inhibitors—lisinopril (4.90, 3.46–6.94), ramipril (13.97, 10.73–18.19), perindopril (7.79, 3.24–18.74); and angiotensin II receptor blockers—candesartan (4.82, 2.41–9.65), valsartan (3.28, 2.06–5.21). Immunomodulating agents (14 drugs; 185 cases, 12.84%) showed significant disproportionality for platinum compounds (bleomycin, 29.21, 15.67–54.44), antimetabolites—gemcitabine (7.13, 5.26–9.66), hydroxycarbamide (4.64, 1.74–12.38); immunosuppressants—belimumab (29.21, 15.67–54.44), voclosporin (6.65, 2.98–14.81), leflunomide (3.69, 2.18–6.23); and interferons—interferon alfa-2a (24.05, 9.00–64.30), alfa-2b (9.01, 4.29–18.93), peginterferon alfa-2b (5.70, 3.49–9.32), beta-1b (3.91, 2.54–6.00). Additional signals emerged in musculoskeletal, hematological, anti-infective, alimentary/metabolic, and other therapeutic categories.

Fig. 3.

Forest plot of drugs with positive signals (top 50 drugs by frequency).

Table 2.

Four signal values of drugs with positive signals associated with Raynaud’s phenomenon.

First-level classification of ATC	Drug	Cases	ROR (95% Cl)	PRR (Chi-square value)	EBGM (EBGM05)	IC (IC025)
Various	Gadoteridol	3	13.78 (4.43–42.82)	13.74 (35.42)	13.73 (5.32)	3.78 (2.11)
Nervous	Amfetamine	5	7.71 (3.21–18.55)	7.7 (29.12)	7.69 (3.69)	2.94 (1.28)
Nervous	Amfetamine; Dexamfetamine	20	5.92 (3.81–9.18)	5.91 (81.22)	5.89 (4.08)	2.56 (0.89)
Nervous	Atogepant	4	5.11 (1.91–13.62)	5.1 (13.18)	5.1 (2.24)	2.35 (0.68)
Nervous	Atomoxetine	44	10.54 (7.83–14.19)	10.51 (375.02)	10.42 (8.12)	3.38 (1.71)
Nervous	Cabergoline	17	24.43 (15.15–39.39)	24.29 (378.3)	24.2 (16.23)	4.6 (2.93)
Nervous	Carbidopa; entacapone; Levodopa	6	7.24 (3.25–16.13)	7.22 (32.14)	7.22 (3.69)	2.85 (1.18)
Nervous	Dexmethylphenidate	6	10.64 (4.77–23.71)	10.61 (52.17)	10.6 (5.42)	3.41 (1.74)
Nervous	Eletriptan	6	7.77 (3.49–17.31)	7.76 (35.26)	7.75 (3.96)	2.95 (1.29)
Nervous	Erenumab	55	5.22 (4–6.81)	5.21 (184.82)	5.16 (4.13)	2.37 (0.7)
Nervous	Ethosuximide	4	30.5 (11.4–81.58)	30.28 (113.18)	30.26 (13.28)	4.92 (3.25)
Nervous	Fremanezumab	15	8.34 (5.02–13.85)	8.33 (96.37)	8.3 (5.43)	3.05 (1.39)
Nervous	Galcanezumab	42	7.78 (5.74–10.55)	7.77 (245.42)	7.7 (5.97)	2.95 (1.28)
Nervous	Lisdexamfetamine	100	25.19 (20.65–30.74)	25.05 (2255.84)	24.49 (20.74)	4.61 (2.95)
Nervous	Methylphenidate	70	8.28 (6.54–10.49)	8.27 (439.96)	8.15 (6.69)	3.03 (1.36)
Nervous	Milnacipran	4	7.06 (2.65–18.84)	7.05 (20.75)	7.04 (3.1)	2.82 (1.15)
Nervous	Oxybate sodium	172	11.36 (9.75–13.23)	11.33 (1555.77)	10.92 (9.61)	3.45 (1.78)
Nervous	Rimegepant	8	4.68 (2.34–9.37)	4.68 (23.1)	4.67 (2.61)	2.22 (0.56)
Nervous	Rizatriptan	9	18.82 (9.77–36.24)	18.74 (150.84)	18.7 (10.81)	4.23 (2.56)
Nervous	Solriamfetol	6	14.25 (6.39–31.78)	14.21 (73.57)	14.19 (7.25)	3.83 (2.16)
Nervous	Sumatriptan	35	6.89 (4.94–9.61)	6.88 (174.48)	6.83 (5.17)	2.77 (1.11)
Nervous	Zonisamide	4	5.27 (1.98–14.06)	5.27 (13.82)	5.26 (2.32)	2.4 (0.73)
Musculo-skeletal	Alendronic acid	80	10.14 (8.12–12.65)	10.11 (644.92)	9.94 (8.26)	3.31 (1.65)
Musculo-skeletal	Collagenase clostridium histolyticum	6	4.3 (1.93–9.59)	4.3 (15.18)	4.3 (2.2)	2.1 (0.44)
Musculo-skeletal	Pamidronic acid	6	6.62 (2.97–14.76)	6.61 (28.56)	6.61 (3.38)	2.72 (1.06)
Musculo-skeletal	Rofecoxib	35	4.12 (2.95–5.74)	4.12 (81.88)	4.09 (3.1)	2.03 (0.37)
Immunomodulating	Anastrozole	14	4.32 (2.55–7.3)	4.32 (35.55)	4.3 (2.77)	2.11 (0.44)
Immunomodulating	Avapritinib	8	4.77 (2.38–9.54)	4.76 (23.75)	4.76 (2.66)	2.25 (0.58)
Immunomodulating	Belimumab	23	4.51 (2.99–6.8)	4.51 (62.43)	4.49 (3.18)	2.17 (0.5)
Immunomodulating	Bicalutamide	5	6.17 (2.56–14.83)	6.16 (21.59)	6.15 (2.95)	2.62 (0.95)
Immunomodulating	Bleomycin	10	29.21 (15.67–54.44)	29.01 (269.9)	28.95 (17.19)	4.86 (3.19)
Immunomodulating	Gemcitabine	42	7.13 (5.26–9.66)	7.12 (218.78)	7.06 (5.47)	2.82 (1.15)
Immunomodulating	Hydroxycarbamide	4	4.64 (1.74–12.38)	4.64 (11.4)	4.63 (2.04)	2.21 (0.54)
Immunomodulating	Interferon alfa-2a	4	24.05 (9–64.3)	23.92 (87.8)	23.9 (10.5)	4.58 (2.91)
Immunomodulating	Interferon alfa-2b	7	9.01 (4.29–18.93)	8.99 (49.67)	8.98 (4.83)	3.17 (1.5)
Immunomodulating	Interferon beta-1b	21	3.91 (2.54–6)	3.9 (45.17)	3.89 (2.72)	1.96 (0.29)
Immunomodulating	Leflunomide	14	3.69 (2.18–6.23)	3.68 (27.29)	3.67 (2.37)	1.88 (0.21)
Immunomodulating	Peginterferon alfa-2b	16	5.7 (3.49–9.32)	5.7 (61.76)	5.68 (3.77)	2.51 (0.84)
Immunomodulating	Ponatinib	11	5.29 (2.93–9.56)	5.28 (38.1)	5.27 (3.21)	2.4 (0.73)
Immunomodulating	Voclosporin	6	6.65 (2.98–14.81)	6.64 (28.7)	6.63 (3.39)	2.73 (1.06)
Genitourinary	Ethinylestradiol; etonogestrel	15	3.39 (2.04–5.63)	3.39 (25.14)	3.38 (2.21)	1.76 (0.09)
Genitourinary	Raloxifene	7	4.05 (1.93–8.5)	4.05 (16.03)	4.04 (2.17)	2.01 (0.35)
Cardiovascular	Atenolol	9	4.2 (2.18–8.08)	4.2 (21.88)	4.19 (2.42)	2.07 (0.4)
Cardiovascular	Bisoprolol	56	22.13 (16.99–28.83)	22.02 (1109.44)	21.75 (17.43)	4.44 (2.78)
Cardiovascular	Bosentan	43	4.65 (3.44–6.28)	4.64 (121.73)	4.61 (3.58)	2.2 (0.54)
Cardiovascular	Candesartan	8	4.82 (2.41–9.65)	4.82 (24.17)	4.81 (2.69)	2.27 (0.6)
Cardiovascular	Carteolol	4	69.47 (25.85–186.69)	68.35 (265.29)	68.29 (29.86)	6.09 (4.42)
Cardiovascular	Fluvastatin	3	9.12 (2.94–28.31)	9.1 (21.62)	9.09 (3.52)	3.18 (1.52)
Cardiovascular	Guanfacine	15	22.97 (13.82–38.19)	22.85 (312.39)	22.77 (14.88)	4.51 (2.84)
Cardiovascular	Hydrochlorothiazide	9	4.86 (2.53–9.35)	4.85 (27.5)	4.85 (2.8)	2.28 (0.61)
Cardiovascular	Labetalol	15	55.95 (33.59–93.2)	55.23 (796.06)	55.04 (35.91)	5.78 (4.11)
Cardiovascular	Lisinopril	32	4.9 (3.46–6.94)	4.9 (98.49)	4.87 (3.64)	2.28 (0.62)
Cardiovascular	Macitentan	38	4.37 (3.17–6.01)	4.36 (97.6)	4.33 (3.32)	2.11 (0.45)
Cardiovascular	Metoprolol	21	3.56 (2.32–5.46)	3.55 (38.38)	3.54 (2.47)	1.82 (0.16)
Cardiovascular	Nadolol	3	18.91 (6.08–58.79)	18.83 (50.61)	18.81 (7.28)	4.23 (2.56)
Cardiovascular	Nebivolol	11	11.69 (6.46–21.14)	11.66 (106.93)	11.63 (7.08)	3.54 (1.87)
Cardiovascular	Nifedipine	15	12.57 (7.57–20.89)	12.54 (158.76)	12.5 (8.17)	3.64 (1.98)
Cardiovascular	Perindopril	5	7.79 (3.24–18.74)	7.78 (29.5)	7.77 (3.73)	2.96 (1.29)
Cardiovascular	Propranolol	20	8.04 (5.18–12.49)	8.03 (122.55)	8 (5.54)	3 (1.33)
Cardiovascular	Ramipril	56	13.97 (10.73–18.19)	13.92 (663.14)	13.76 (11.03)	3.78 (2.12)
Cardiovascular	Simvastatin	22	3.43 (2.26–5.21)	3.43 (37.63)	3.41 (2.4)	1.77 (0.11)
Cardiovascular	Valsartan	18	3.28 (2.06–5.21)	3.28 (28.35)	3.27 (2.22)	1.71 (0.04)
Blood	Epoprostenol	11	3.79 (2.1–6.85)	3.79 (22.53)	3.78 (2.3)	1.92 (0.25)
Blood	Selexipag	12	4.72 (2.68–8.32)	4.71 (35.03)	4.7 (2.93)	2.23 (0.57)
Antiinfectives	Ciprofloxacin	29	3.35 (2.32–4.82)	3.34 (47.35)	3.33 (2.45)	1.73 (0.07)
Antiinfectives	Trimethoprim	3	7.77 (2.5–24.12)	7.76 (17.64)	7.75 (3)	2.95 (1.29)
Alimentary and metabolism	Asfotase alfa	7	4.18 (1.99–8.77)	4.17 (16.86)	4.17 (2.24)	2.06 (0.39)
Alimentary and metabolism	Sulfasalazine	7	4 (1.91–8.4)	4 (15.72)	3.99 (2.15)	2 (0.33)

Given the significant differences in expertise, reporting motivations, and content quality between non-healthcare and healthcare reporters, this study conducted stratified disproportionality analyses based on reporter backgrounds. Reports from healthcare professionals are more reliable in pharmacovigilance. Forest plots of the top 30 drugs associated with RP, stratified by reporter type (non-healthcare vs. healthcare), are presented in Fig. 4A and B. Among non-healthcare reporters, 48 drugs were identified as positive signals, while 53 drugs were flagged among healthcare professionals. Among pharmacovigilance signals detected in the healthcare professional subgroup, 86.79% overlapped with signals from the overall background population. Key emerging signals comprised amlodipine/valsartan, atorvastatin, dorzolamide/timolol, interferon beta-1a, peginterferon alfa-2a, sildenafil, and teriflunomide. Among the 48 positive signal drugs detected in the non-healthcare reporter subgroup, 68.75% demonstrated overlap with signals from the overall background population. This high concordance rate across reporter subgroups confirms the robustness of our findings. Additionally, healthcare professionals reported a higher proportion of RP cases compared to non-healthcare reporters, with a ratio of 1.58:1 in total case numbers (see Supplementary Table S1 for Consumer population reporting and Healthcare population reporting for details).

Fig. 4.

Forest plots of positive signal drugs in stratified populations (top 30 drugs by case count): (A) Positive signal drugs for RP reported by consumers. (B) Positive signal drugs for RP reported by healthcare professionals. (C) RP-associated positive signal drugs in the 18–64-year-old population.

Within the FAERS database, patients across all age groups were reported to experience RP. As demonstrated in the population characteristics, RP cases were predominantly observed in the 25–80-year age group, comprising 2413 cases (87.08%). To minimize bias from physiological variations in extreme age groups, a stratified analysis was conducted focusing on the 18–64-year age group, identifying 44 positive signal drugs. A forest plot of the top 30 drugs by frequency is presented in Fig. 4C, with 86.36% of these drugs overlapping with the positive signals identified in the overall analysis. Despite potential confounding factors in younger or elderly populations, the positive signals derived from the overall analysis remained robust. Notably, certain drugs consistently showed positive signals across all stratified analyses, including alendronic acid, amfetamine, dexamfetamine, atomoxetine, bisoprolol, eletriptan, erenumab, fremanezumab, galcanezumab, and others (see Supplementary Table S1 for the age group of 18–64 for details)

TTO analysis

In the FAERS database, the TTO of adverse reactions is often incomplete. This study analyzed cases with both RP and valid records of medication initiation and adverse event onset. Figure 5 displays the cumulative time-to-onset curve, showing a rapid increase in RP cases within the first 50 days after medication initiation, followed by a stabilized risk over time without significant anomalies. Among the 68 positive signal drugs, 12 had ≥ 10 reported RP cases. The median TTO and IQR were calculated for each drug, along with Weibull shape parameter (WSP) testing (Table 3). Ciprofloxacin exhibited the shortest median TTO of 44 days (IQR 1–17), while interferon beta-1b showed the longest median TTO of 579 days (IQR 268.5–3363). All 12 drugs demonstrated WSP values consistent with a late failure pattern, indicating an increasing risk of RP occurrence over prolonged exposure.

Fig. 5.

Cumulative time-to-onset curve of adverse reactions involving Raynaud’s phenomenon.

Table 3.

Median time-to-onset of adverse reactions and WSP test types for positive signal drugs with ≥ 10 cases.

Drug	n	TTO	Scale parameter: α (95% CI)	Shape parameter: β (95% CI)	Type
Alendronic acid	27	124 (1, 354)	0.4799 (0.3517, 0.6548)	152.6712 (66.7735, 349.0684)	Late failure
Anastrozole	12	24 (24, 45.5)	0.6803 (0.4581, 1.0101)	84.0239 (34.5968, 204.0651)	Late failure
Atomoxetine	16	163 (31.5, 173)	0.5906 (0.4004, 0.8710)	180.8522 (75.6595, 432.2991)	Late failure
Bosentan	16	272 (61.5, 666)	0.5630 (0.3849, 0.8234)	526.9169 (210.1516, 1321.1481)	Late failure
Ciprofloxacin	19	4 (1, 17)	0.9180 (0.6451, 1.3064)	7.7198 (4.5947, 12.9707)	Late failure
Erenumab	14	93 (22.25, 177.75)	0.6371 (0.4151, 0.9780)	117.5616 (49.5911, 278.6939)	Late failure
Galcanezumab	13	20 (3, 33)	0.5921 (0.3978, 0.8813)	30.9749 (11.6934, 82.0506)	Late failure
Interferon beta-1b	11	579 (268.5, 3363)	0.5692 (0.3660, 0.8854)	2350.0765 (781.4487, 7067.4629)	Late failure
Lisdexamfetamine	13	74 (19, 184)	0.7068 (0.4604, 1.0850)	141.1263 (62.7279, 317.5083)	Late failure
Lisinopril	25	93 (1, 93)	0.4407 (0.3221, 0.6029)	64.8371 (25.3236, 166.0054)	Late failure
Macitentan	11	144 (42.5, 350)	0.7633 (0.4854, 1.2003	280.7107 (123.5528, 637.7716)	Late failure
Methylphenidate	10	26 (12, 869.5)	0.4082 (0.2497, 0.6671)	220.5357 (44.2235, 1099.7776)	Late failure

Discussion

This study represents the first systematic pharmacovigilance analysis of drug-induced RP spanning 20 years using the FAERS database. Through descriptive analysis and multidimensional disproportionality assessments, we identified population characteristics and risk signals associated with medication-related RP. Notably, among the annual cases of RP reported, the number documented in 2004 was significantly lower than in subsequent years. This finding should not be directly interpreted as indicative of a lower incidence in 2004; rather, it is more likely attributable to limitations in the pharmacovigilance infrastructure at that time. This phenomenon and its contextual background are associated with factors including the absence of a robust adverse drug reaction (ADR) reporting system, lack of standardized reporting procedures across healthcare institutions, limited public awareness of ADRs, and insufficient understanding among healthcare professionals regarding reporting obligations. Nevertheless, the annual reported cases of RP overall demonstrated an increasing trend. In traditional frequency-based analysis, immunomodulators ranked highest (21 out of 50 drugs). However, after multi-algorithm validation, nervous system agents (21 drugs, 632 cases) and cardiovascular drugs (20 drugs, 403 cases) emerged as the dominant categories among the 68 positive signal drugs, exhibiting both greater drug diversity and higher RP case numbers. Notably, while immunomodulators were frequently implicated, their proportion of positive signals (14 drugs, 185 cases) was significantly lower than that of nervous/cardiovascular agents. Stratified analyses confirmed the robustness of the association between drug exposure and Raynaud’s phenomenon. Based on reports submitted by individuals with a medical or healthcare background, 53 positive drugs were identified, with a signal overlap rate of 86.79% compared to the overall dataset. The proportion of reports from healthcare professionals was significantly higher than those from non-professionals (ratio of 1.58:1), and these professional reports were considered to have greater reliability. Furthermore, among individuals aged 18–64 years, 44 positive drugs were identified, 86.36% of which overlapped with the overall signal, suggesting that age stratification effectively excluded the influence of physiological extremes. Notably, several drugs—such as alendronic acid, amfetamine/dexamfetamine, and atomoxetine—consistently exhibited positive signals across different stratification backgrounds.

The findings of this study are consistent with the conclusions drawn by Khouri et al.¹⁰ in their review regarding the association between medications and RP. Drug classes highlighted in that review, including β-blockers, ergot alkaloids, dopaminergic agonists, sympathomimetics, cyclosporines, chemotherapeutic agents, interferons, and tyrosine kinase inhibitors, all yielded positive signals in the present analysis. Similarly, Lee et al.¹², who also investigated the association between CGRP inhibitors and RP using FAERS data, reported results aligning with our findings (specific ROR [95% CI] comparisons: Atogepant: 23.45 [7.54–73.01] vs. this study 5.11 [1.91–13.62]; Erenumab: 12.58 [9.12–17.36] vs. 5.22 [4.00–6.81]; Fremanezumab: 32.93 [19.06–56.91] vs. 8.34 [5.02–13.85]; Galcanezumab: 31.38 [21.43–45.94] vs. 7.78 [5.74–10.55]; Rimegepant: 12.87 [5.34–31.01] vs. 4.68 [2.34–9.37]; Sumatriptan: 21.39 [12.40–36.91] vs. 6.89 [4.94–9.61]; Rizatriptan: 10.66 [4.00–28.44] vs. 18.82 [9.77–36.24]; Eletriptan: 17.02 [8.09–35.75] vs. 7.77 [3.49–17.31]). Although the specific reported ROR values differ—potentially attributable to variations in sample size, scope of investigated drugs, or data cleaning methodologies—both studies are consistent in suggesting an association between the aforementioned drugs and RP. Furthermore, the analysis by Gérard et al.¹³ using the WHO pharmacovigilance database on the association of CGRP-targeting drugs with RP also reported partial overlap with the positive signal drugs identified in our study; for instance, erenumab, galcanezumab, and fremanezumab all demonstrated positive associations with RP. Putkaradze et al.¹⁴, through a pattern recognition study of adverse events within the WHO pharmacovigilance database, specifically identified drugs potentially inducing RP (e.g., β-adrenergic receptor blockers, sympathomimetics and stimulants, interferons, dopaminergic agonists). These findings are further corroborated by the positive signals observed for these drug classes in the present study. Collectively, the results from these aforementioned studies support the robustness of our findings.

The 68 positive signal drugs identified in this study span eight ATC categories, reflecting the complex pathophysiological mechanisms underlying drug-induced RP. Nervous system agents, which constituted the largest category (21 drugs, n = 632), primarily exert RP risk through dysregulation of sympathetic nervous activity and vasomotor function¹⁵. These drugs may enhance sympathetic excitability, promote norepinephrine release, and activate peripheral vascular α1-adrenergic receptors, leading to abnormal vasoconstriction¹⁶. Some agents may also interfere with serotonin or dopamine systems¹⁷, exacerbating vascular endothelial dysfunction or inhibiting nitric oxide-mediated vasodilation, thereby triggering RP. Prolonged use of these drugs can increase vascular smooth muscle sensitivity to vasoconstrictive substances, further aggravating peripheral vasospasm. For example, CGRP inhibitors such as erenumab (n = 55, ROR 5.22) and fremanezumab (n = 15, ROR 8.34) block CGRP-receptor binding, suppressing its potent vasodilatory effects. This results in elevated peripheral vascular tone, particularly under cold exposure or stress, predisposing individuals to RP. Similar mechanisms have been reported in prior studies^13,18. Central nervous system stimulants such as lisdexamfetamine (n = 100, ROR 25.19) and methylphenidate (n = 70, ROR 8.28) primarily induce Raynaud’s phenomenon (RP) by inhibiting dopamine and norepinephrine reuptake^19,20. This mechanism significantly enhances sympathetic nervous output, activating vascular smooth muscle α1-adrenergic receptors and triggering peripheral arteriolar spasm. Additionally, these drugs may downregulate endothelial nitric oxide synthase (eNOS) expression, reducing nitric oxide (NO)-mediated vasodilation and exacerbating vasoconstriction²¹, which are established risk factors for RP. For triptans (e.g., sumatriptan, eletriptan, rizatriptan), existing literature highlights their potential to induce RP due to their potent vasoconstrictive properties^12,22. Notably, oxybate sodium (n = 172, ROR 11.36) accounted for a substantial number of RP cases, with RP representing 0.26% of all adverse reactions reported for this drug. As a GABA_B receptor agonist, oxybate sodium exerts sedative and sleep-regulating effects, potentially modulating neurotransmitter balance in the central nervous system and indirectly influencing sympathetic activity. While no studies have directly linked oxybate sodium to RP, the literature suggests that drugs with similar mechanisms may increase cardiovascular risk^23–25. Further investigation—including large-scale epidemiological investigations, case–control studies, and mechanistic research—is required to clarify the potential causal relationship between oxybate sodium and RP and to explore its underlying pathways.

Cardiovascular drugs (20 types, n = 403) have been linked to RP, which is primarily associated with their effects on vasomotor function. This study revealed class heterogeneity in the mechanisms of RP induction by cardiovascular drugs. Non-selective-blockers such as Propranolol, Carteolol, and Nadolol may be more likely to cause vasoconstriction by blocking 2 receptor-mediated vasodilation while enhancing 1 receptor-mediated vasoconstrictive effects, significantly increasing the risk of peripheral vasospasm and subsequently triggering RP. This mechanistic category has long been recognized as the primary cause of drug-induced RP^10,21. However, selective 1-blockers, ACE inhibitors, ARBs, calcium channel blockers, and endothelin receptor antagonists (including Atenolol, Bisoprolol, Metoprolol, Lisinopril, Perindopril, Ramipril, Candesartan, Valsartan, Nifedipine, Bosentan, and Macitentan) also showed positive signals in this study. Pharmacologically, these drugs are considered low-risk for inducing RP due to their vasodilatory or blood flow-improving effects. Nevertheless, clinical case reports still exist, particularly in patients with pre-existing peripheral vascular diseases.

Immunomodulators (14 types, n = 185) are closely associated with RP risk through direct vascular endothelial injury and immune-mediated microcirculatory dysfunction. In this study, interferon drugs such as Interferon alfa-2a, Interferon alfa-2b, Interferon beta-1b, and Peginterferon alfa-2b, along with antitumor/immunomodulatory agents like Hydroxycarbamide, Bleomycin, and Ponatinib, were identified as clear risk factors for RP. These drugs may induce endothelial cell dysfunction, disrupting the balance between vasodilation and vasoconstriction. Prolonged use can lead to vascular structural abnormalities, such as endothelial damage and occlusive microangiopathy, which further impair normal vascular function and increase RP risk^5,21,26,27. The antimetabolite Gemcitabine (n = 42, ROR 7.13) triggers ischemic vasospasm by inducing reactive oxygen species (ROS) bursts, causing mitochondrial dysfunction and DNA damage in endothelial cells, while upregulating von Willebrand factor (vWF) release to promote platelet adhesion and microthrombosis. Belimumab (n = 23, ROR 4.51), a B-cell activating factor (BAFF) inhibitor, is a therapeutic agent for systemic lupus erythematosus (SLE)²⁸. Although Belimumab demonstrates favorable safety and efficacy in SLE and systemic sclerosis²⁹, no direct evidence currently links it to the induction or exacerbation of RP. However, given the high prevalence of RP in autoimmune diseases like systemic sclerosis and Belimumab’s broad immunomodulatory effects, this association may relate to the underlying disease (e.g., SLE) itself.

In this study, although other categories of drugs accounted for a small proportion, such as antibacterial drugs like Trimethoprim, Sulfasalazine, and Ciprofloxacin, which also showed positive signals, these classes are considered uncommon triggers of RP in terms of literature reports and pharmacological mechanisms. Some of the positive signal drugs identified in this study align with existing evidence-based medical evidence on high-risk factors for RP, further enhancing the reliability of the study results. Examples include CGRP inhibitors, triptans, non-selective β-blockers, interferon-based drugs, and antitumor/immunomodulatory agents. However, for some drugs with positive signals in this study for which there is currently limited supporting evidence in the literature, further research is needed to validate their associations.

Limitations

While this study employed a multi-algorithm framework (ROR/PRR/BCPNN/MGPS) to mitigate the limitations of individual methods, inherent shortcomings of the FAERS database must be acknowledged. These include underreporting, duplicate entries, and confounding factors (e.g., polypharmacy, comorbidities) that may compromise the accuracy of risk signals^30,31. Disproportionality analyses only suggest statistical associations and cannot establish causality³². Furthermore, the overrepresentation of cases from Western countries and female populations (gender ratio 3.13:1) may limit the generalizability of conclusions³³. For instance, certain cardiovascular drugs—such as selective β1-blockers, ACE inhibitors, ARBs, calcium channel blockers, and endothelin receptor antagonists—are clinically used to improve cardiovascular outcomes or alleviate RP symptoms³⁴. Despite showing positive signals in this study, these results do not confirm causal relationships and may reflect confounding by patient-specific factors (e.g., comorbidities, lifestyle, drug interactions). Additional limitations include incomplete time-to-onset records, unresolved long-term risk dynamics, and challenges in disentangling drug effects from underlying conditions (e.g., rheumatoid arthritis). Although stratified analyses validated signal robustness, potential modifiers such as dosage, treatment duration, and genetic background were not assessed. Future research should integrate real-world data, experimental studies, and multidisciplinary approaches to validate mechanisms and refine drug safety evaluation frameworks.

Conclusions

This study identified 68 positive signal drugs significantly associated with RP based on FAERS data. Nervous system agents, cardiovascular drugs, and immunomodulators exhibited higher diversity, frequency, and risk levels compared to other drug categories. Clinicians should exercise heightened vigilance for RP risk when prescribing CGRP inhibitors, triptans, non-selective beta-blockers, interferon-based therapies, and anticancer/immunomodulatory agents, particularly in middle-aged and elderly women and patients with rheumatoid arthritis, who may be more susceptible to drug-induced RP. These findings highlight the significance of integrating pharmacovigilance insights into clinical decision-making to mitigate adverse vascular events.

Methods

Data acquisition

This study was conducted following the READUS-PV guidelines³². The FAERS database contains publicly accessible, anonymized patient data; thus, ethical committee approval was not required for its use. The data for this study encompass adverse event reports spanning from the first quarter of 2004 to the second quarter of 2024. ASCII data files for the United States were downloaded from the FDA Adverse Event Reporting System (FAERS) database at https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html. Quarterly data files for each year were downloaded and saved, resulting in 82 compressed quarterly files stored on a local disk. These files comprise the following data tables: Demographic and Administrative Information (DEMO), Drug Information (DRUG), Adverse Events (REAC), Patient Outcomes (OUTC), Report Sources (RPSR), Drug Therapy Start and End Dates (THER), and Indications for Use/Diagnosis (INDI).

Data cleaning and filtering

Data cleaning

All data analyses in this study were performed using R software (version 4.3.3). The data processing phase aimed to systematically clean and map raw data obtained from the FAERS database to ensure quality, completeness, and consistency. The workflow is illustrated in Fig. 6. Deduplication: For duplicate records sharing the same CASEID (a unique identifier for FAERS reports), the entry with the latest FDA_DT (date the FDA received the report) was retained. When both CASEID and FDA_DT were identical, the record with the higher PRIMARYID (a unique numerical identifier for FAERS reports) was selected. This step ensured that each report represented the most recent and unique entry, minimizing potential bias from duplicate data. In the DRUG file, ROLE_COD (the role code indicating the drug’s association with the adverse event) was used to identify drug involvement. The possible role codes include: PS (Primary Suspect: drug considered most likely to cause the adverse event); SS (Secondary Suspect: drug potentially contributing to the event); C (Concomitant: drug administered concurrently without established causal relationship); I (Interacting: drug involved in drug-drug interactions leading to the event); OT (Other: unclassifiable circumstances). This study exclusively included reports with ROLE_COD designated as “PS” (Primary Suspect) to focus analysis on medications most likely implicated in adverse events. This selection process enhances data relevance and analytical accuracy^35,36.

Fig. 6.

Data processing workflow.

Data mapping

Drug name standardization is critical in this study, serving as a key step to ensure data quality and analytical accuracy. Non-standardized drug names are a common issue in drug safety monitoring and adverse event reporting, potentially leading to duplication, omission, or confusion, thereby compromising result reliability. The DiAna Dictionary was utilized for drug name mapping, converting nearly 99% of drug names to their active ingredients with high precision—significantly outperforming other tools such as RxNorm (76.32%). This approach markedly enhanced data processing efficiency and accuracy. During standardization, original drug names were retained for subsequent in-depth analysis and validation, ensuring data integrity and traceability³⁷. Standardized data are stored in an open-access repository: https://osf.io/zqu89/.

The active ingredient names provided by the DiAna Dictionary were mapped to the DRUGNAME column in the DRUG table. Non-standardized drug names in the DRUGNAME column were redefined to their correct active ingredient names. This standardization process ensured consistency and accuracy in drug nomenclature, thereby mitigating data errors caused by naming discrepancies. Cases reporting “Raynaud’s Phenomenon” were identified from the REAC data file. Using the PRIMARYID, the DRUG, DEMO, OUTC, THER, and INDI files were cross-mapped to integrate multidimensional data, including patient demographics, drug details, adverse event descriptions, clinical outcomes, therapy timelines, and diagnostic indications. This linkage created a comprehensive analysis dataset. From the integrated dataset, drugs with ≥ 3 RP reports were retained to ensure the statistical power and reliability of the analysis³⁸.

Data analysis

Fundamental analysis

This study employed disproportionality analysis to evaluate potential associations between drugs and RP events. Disproportionality analysis is a statistical method based on 2 × 2 contingency tables, which identifies significant association signals by calculating the reporting ratios of drugs and adverse events. The following four methods were used to comprehensively assess drug-RP associations³⁸. Reporting odds ratio (ROR): This method evaluates associations by comparing the reporting proportion of a specific drug and adverse event. ROR is intuitive and suitable for large-scale pharmacovigilance data analysis, effectively identifying potential drug safety signals. Proportional reporting ratio (PRR): Similar to ROR but with a more direct calculation, PRR assesses associations by comparing the reporting proportion of a specific drug-adverse event pair to other drugs. Widely used in pharmacovigilance, it offers high sensitivity. Bayesian confidence interval neural network (BCINN): A Bayesian statistical approach that constructs complex models to evaluate drug-event associations. This method accounts for data uncertainty, provides robust results, and is an advanced analytical tool in pharmacovigilance. Multi-item gamma-Poisson shrinkage (MGPS): This method identifies disproportionality by fitting a multi-distribution model to count data, effectively detecting imbalances between drugs and adverse events. The combined use of these four disproportionality analysis methods leverages their strengths, mitigates limitations of single approaches, and enhances flexibility and adaptability for complex scenarios with diverse data characteristics and analytical requirements. This study classified a drug as a positive signal for RP only if all four methods met their respective thresholds³⁹. Formulas and thresholds are detailed in Table 4. Drugs yielding positive signals were categorized according to the Anatomical Therapeutic Chemical (ATC) classification system (accessed via https://atcddd.fhi.no/atc_ddd_index/) to facilitate analysis of association patterns between drug classes and falls.

Table 4.

Formulas and thresholds for disproportionality analysis.

Algorithms	Formulas	Thresholds
ROR		ROR (Lower 95% CI) > 1, N ≥ 2
PRR		PRR ≥ 2, χ2 ≥ 4, N ≥ 3
BCPNN		IC (Lower 95% CI) ≥ 0
MGPS		EBGM (Lower 95% CI) ≥ 2 N ≥ 0

In addition, this study conducted descriptive analyses, including characteristics of the RP population (e.g., age, gender, outcomes, and patient indications), the demographic and regional distribution of RP case reports, and the time-to-onset (TTO) of adverse reactions. TTO refers to the interval between the initiation of drug use and the occurrence of an adverse event, serving as a critical indicator for evaluating drug safety. To describe TTO, this study utilized the median and interquartile range (IQR) as key statistical measures to reflect its distribution characteristics. Furthermore, the Weibull shape parameter (WSP) test was applied to explore the temporal patterns of TTO distribution. Analyzing the shape of TTO distributions helps reveal the underlying patterns of drug-related adverse event occurrence⁴⁰.

Secondary analysis

The secondary analysis aimed to validate the robustness and reliability of the fundamental analysis results. A stratified disproportionality analysis was conducted comparing the 18–64-year-old population to the overall population. Age is a critical influencing factor for RP, with the 18–65-year age group typically representing the high-risk demographic for this adverse event⁴¹. By restricting the analysis to the 18–65-year age range, potential biases introduced by other age groups (e.g., children or the elderly) could be minimized. While elderly populations may be more prone to RP due to factors such as polypharmacy or underlying comorbidities, the analysis of the 18–65-year age group better reflects drug-specific safety concerns.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Pingping Zheng: Contributed to the development of methods, data collection, data analysis, and manuscript drafting; Xiulian Zheng: Contributed to data collection and data analysis; Chen Qun: Contributed to data collection and data analysis; Dongmei Wang: Contributed to data analysis; Longzhuan Huang: Contributed to data analysis; Jianping Lin: Provided financial support, supervised the methodology, and revised the manuscript; Shaoqing Chen: Provided financial support, supervised the methodology, and revised the manuscript. All authors contributed to manuscript drafting and revision, and approved the final version.

Data availability

All data generated and analyzed in this study are provided within the manuscript or supplementary information files.

Declarations

Competing interests

The authors declare no competing interests.