
==== Front
BMC Med
BMC Med
BMC Medicine
1741-7015
BioMed Central London

39285302
3626
10.1186/s12916-024-03626-2
Research
Real-world impact of acupuncture on analgesics and healthcare resource utilization in breast cancer survivors with pain
Ng Ding Quan 12
Lee Sanghoon 3
Lee Richard T. 4
Wang Yun yunwang@chapman.edu

2
Chan Alexandre a.chan@uci.edu

1
1 https://ror.org/04gyf1771 grid.266093.8 0000 0001 0668 7243 School of Pharmacy & Pharmaceutical Sciences, University of California Irvine, 802 W Peltason Dr, Irvine, CA 92697-4625 USA
2 https://ror.org/0452jzg20 grid.254024.5 0000 0000 9006 1798 School of Pharmacy, Chapman University, RK 94-206, 9401 Jeronimo Road, Irvine, CA 92618 USA
3 https://ror.org/01zqcg218 grid.289247.2 0000 0001 2171 7818 College of Korean Medicine, Kyung Hee University, Seoul, South Korea
4 grid.410425.6 0000 0004 0421 8357 Integrative Medicine Program, Departments of Supportive Care Medicine and Medical Oncology, City of Hope Comprehensive Cancer Center, Duarte, CA USA
16 9 2024
16 9 2024
2024
22 39416 4 2024
10 9 2024
© The Author(s) 2024
2024
https://creativecommons.org/licenses/by/4.0/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Background

This study evaluated the real-world impact of acupuncture on analgesics and healthcare resource utilization among breast cancer survivors.

Methods

From a United States (US) commercial claims database (25% random sample of IQVIA PharMetrics® Plus for Academics), we selected 18–63 years old malignant breast cancer survivors experiencing pain and ≥ 1 year removed from cancer diagnosis. Using the difference-in-difference technique, annualized changes in analgesics [prevalence, rates of short-term (< 30-day supply) and long-term (≥ 30-day supply) prescription fills] and healthcare resource utilization (healthcare costs, hospitalizations, and emergency department visits) were compared between acupuncture-treated and non-treated patients.

Results

Among 495 (3%) acupuncture-treated patients (median age: 55 years, stage 4: 12%, average 2.5 years post cancer diagnosis), most had commercial health insurance (92%) and experiencing musculoskeletal pain (98%). Twenty-seven percent were receiving antidepressants and 3% completed ≥ 2 long-term prescription fills of opioids. Prevalence of opioid usage reduced from 29 to 19% (P < 0.001) and NSAID usage reduced from 21 to 14% (P = 0.001) post-acupuncture. The relative prevalence of opioid and NSAID use decreased by 20% (P < 0.05) and 19% (P = 0.07), respectively, in the acupuncture-treated group compared to non-treated patients (n = 16,129). However, the reductions were not statistically significant after adjustment for confounding. Patients receiving acupuncture for pain (n = 264, 53%) were found with a relative decrease by 47% and 49% (both P < 0.05) in short-term opioid and NSAID fills compared to those treated for other conditions. High-utilization patients (≥ 10 acupuncture sessions, n = 178, 36%) were observed with a significant reduction in total healthcare costs (P < 0.001) unlike low-utilization patients.

Conclusions

Although adjusted results did not show that patients receiving acupuncture had better outcomes than non-treated patients, exploratory analyses revealed that patients treated specifically for pain used fewer analgesics and those with high acupuncture utilization incurred lower healthcare costs. Further studies are required to examine acupuncture effectiveness in real-world settings.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12916-024-03626-2.

Keywords

Acupuncture
Cancer pain
Real-world evidence
Claims data
Propensity score
Difference-in-difference
issue-copyright-statement© BioMed Central Ltd., part of Springer Nature 2024
==== Body
pmcBackground

Over two million patients are diagnosed with cancer annually in the United States (US), leading to a projected population of 22 million by 2030 [1]. Cancer patients experience poorer quality of life as a consequence of acute and chronic cancer-related symptoms [2, 3], among which pain is highly disabling and arguably the most feared symptom. Prevalence of pain stands at over 40% across cancer types [4, 5], which can increase to over 70% among metastatic cancer patients [4].

In comparison to nonmalignant pain, cancer pain is multifactorial with varying presentations dependent on the cancer type, stage, and cancer treatment received [6]. Aromatase inhibitors, prescribed for reducing breast cancer recurrence in > 50% of all breast cancer cases [7, 8], cause arthralgias in more than half of the patients [9, 10]. Taxane-related agents can induce neuropathic pain that is difficult to treat and debilitating to quality of life [11]. Although opioids and non-steroidal anti-inflammatory drugs (NSAIDs) remain highly relevant in pain management [12, 13], use of these medications is complicated by adverse events (AEs) such as constipation, gastric ulcers, and medication dependence. Optimizing pain control remains an uphill battle in cancer survivorship and supportive care [14–16]. With complex etiologies, significant AEs associated with analgesics use, and improved understanding of psychological and spiritual elements in cancer pain treatment, there has been a paradigm shift towards the integration of conventional pharmacological and complementary non-pharmacological interventions (i.e., integrative medicine) for holistic, safe, and effective pain management [12, 13].

In the recent SIO-ASCO guidelines for pain management, acupuncture received a moderate recommendation, highest among non-pharmacological therapies, for alleviating aromatase inhibitor-induced arthralgias, as well as general and musculoskeletal pain [12]. While published large-scale acupuncture trials have reported significant declines in cancer pain severity [17, 18], it remains unclear if these findings are translatable to real-world practices. The lack of a robust placebo control results in questions regarding the durability of the observed efficacy and criticisms about acupuncture’s identity as a “mega-placebo” [19, 20]. Alongside the dearth of scientific understanding regarding acupuncture among payers, acupuncture utilization is inequitable and largely restricted to patients with private insurance coverage and high disposable income for out-of-pocket expenditure [21]. Novel approaches for assessing the effectiveness and value of acupuncture are needed.

Using a large US commercial claims database, we evaluated the real-world treatment effect of acupuncture for managing cancer pain. Trial-reported benefits of acupuncture for cancer pain are most studied among breast cancer patients who thus serve as the target population for this study [17, 18, 22]. Pain outcomes were determined with analgesics utilization, and we hypothesized that acupuncture treatment had reduced the use of analgesics in comparison to non-treated survivors. The impact on direct medical costs and healthcare resource utilization were also evaluated to investigate the value of acupuncture from the payers’ perspective.

Methods

Data source

This was a retrospective cohort study utilizing a 25% random sample of IQVIA PharMetrics® Plus for Academics (JAN2006-DEC2021), a closed database of fully adjudicated patient-level health plan claims, providing a longitudinal view of inpatient and outpatient services, prescription and office/outpatient administered drugs, costs, and detailed enrollment information. Data contributors are largely commercial health plans, thus making the database representative of the commercially insured US national population for patients under 65 years of age. As data is de-identified in accordance with the US Health Insurance Portability and Accountability Act, institutional review board approval was not required for this study.

Study design

We employed the difference-in-difference (DID) technique, a two-timepoint, pre- and post-intervention analysis, to assess the causal effect of acupuncture exposure with appropriate counterfactual comparisons in a non-randomized setting [23]. In the acupuncture-treated group, the initiation of acupuncture was set as the reference date (i.e., index date) to determine the pre- and post-index periods for comparisons. The index date for non-treated controls was simulated following an observation in this study that patients seek acupuncture after an average of 236 days after a health encounter for pain. The key analysis involved comparing the annualized changes in outcomes, from the pre-index to post-index periods, between acupuncture-treated and non-treated cohorts (Fig. 1).Fig. 1 Illustration of study design and difference-in-difference methodology application

Study population and eligibility criteria

Eligible patients were 18–63 years old (to exclude Medicare-switching), at least 1 year removed from malignant breast cancer diagnosis, continuously enrolled in medical and pharmacy plans for 2 years (from 365-day before to 365-day after index date), and with one or more documented musculoskeletal/general/neoplasm-related pain claims during the 365-day pre-index period. Eligible acupuncture-treated patients first received acupuncture after cancer diagnosis and were not found with acupuncture claims during the pre-index period, while non-treated patients must not have any documented acupuncture-related claims. The algorithms for identifying key health encounters (e.g., acupuncture, breast cancer, and pain) were summarized in Additional file 1: Table S1.

Acupuncture exposure

Acupuncture exposure was determined by the presence of one or more claims for acupuncture using CPT-4 codes after cancer diagnosis (Additional file 1: Table 1). One service unit represents a 15-min treatment duration, and for the purpose of the analyses, an acupuncture session is considered as 30 min long [18].

Outcomes

Analgesics of interest [opioids (non-parenteral), NSAIDs, and adjuvant analgesics such as serotonin-norepinephrine reuptake inhibitors (SNRIs), tricyclic antidepressants (TCAs), and gabapentinoids] were selected based on the NCCN clinical practice guidelines for adult cancer pain [13]. We implemented NDC and HCPCS codes to identify prescription fills (Additional file 1: Table S1). Measures of medication utilization include the prevalence of analgesics users and the rate of short-term (< 30-day supply) and long-term (≥ 30-day supply) prescription fills. The primary outcome is the annualized incremental change in the prevalence of opioid users when compared between acupuncture-treated and non-treated cohorts.

Direct medical costs were determined by the all-cause total healthcare costs paid by both patients and payers. Costs were adjusted to 2022 US dollars using the Consumer Price Index’s annual medical care component [24]. Healthcare resource utilization was defined as the rates of all-cause hospitalizations and emergency department (ED) visits.

Covariates

Sociodemographic characteristics of age, gender, and US geographic region were measured at index date. Healthcare payer (commercial, managed Medicare, Medicare advantage/cost, others) and plan [preferred provider organization (PPO), health maintenance organization (HMO), point-of-service (POS), others] types were determined with monthly enrollment characteristics throughout the 2-year study period. Covariates evaluated during the 365-day pre-index period include total pain-related expenditure, pain subtypes (musculoskeletal, general, or neoplasm-related), Charlson’s comorbidity index excluding cancer-related codes as per National Cancer Institute (NCI) recommendations [25–27], menopausal symptoms, as well as predictors of severe patient-reported symptoms in cancer such as antidepressant use, depression health encounter, and ≥ 2 long-term opioid prescription fills [28]. Cancer-related characteristics (metastases and exposure to anticancer medications) were identified during the period between cancer diagnosis and index date.

Statistical analysis

Primary analysis

Differences between covariates were tested between acupuncture-treated and non-treated patients using Wilcoxon rank-sum test and Pearson’s chi-squared test for continuous and categorical variables, respectively. Distribution of continuous outcomes (cost outcomes) was examined with histogram plots.

For DID analysis, the model was specified as follows: g(Y) = β0 + β1*[post-index] + β2*[acupuncture group] + β3*[post-index × acupuncture group] + ε, with β3 serving as the main coefficient of interest (i.e., DID estimate) for each outcome Y. Models for prevalence of analgesics use as outcomes were modeled with generalized estimating equations (GEE) for estimating population averaged effects, with logit (prevalence < 10%) or log (prevalence ≥ 10%) link function, Huber-White robust standard errors for controlling for heteroskedasticity, first-order autoregression correlation structure for time-varying outcomes, and binomial distribution. Similarly, with GEE and log link function, count outcomes were examined with negative binomial distribution, and cost data implemented gamma distribution. Changes in outcomes from pre- to post-index were estimated using the exponentials of β1 for non-acupuncture patients, and a linear combination of β1 and β3 for acupuncture patients. We further explored the bivariate relationship between predetermined covariates and outcomes that were found to be significant in unadjusted analyses to evaluate potential sources of significant differences not explained by acupuncture exposure.

For covariate adjustment (covariates are listed in Table 1), we implemented an inverse probability of treatment weighting (IPTW) approach as multivariable adjustment with the GEE-DID models could not converge for various outcomes [29]. The probability of treatment assignment (i.e., propensity scores), conditional on the covariates, was estimated with multivariable logistic regression [30]. We employed restricted cubic smoothing splines with five knots to model the relationship between continuous covariates and log-odds of the exposure [31]. Each case was weighted according to the inverse of the propensity score corresponding to its assigned exposure. We stabilized the weights by multiplying them with the marginal probability of the exposure assignment to correct for very large or small weights that could destabilize the estimated effect [31, 32]. Successful covariate balance between the two groups in the weighted cohort was defined as having standardized mean differences achieving <|0.1| for all covariates [33]. Covariate(s) that could not achieve the ideal balance standard were added to the GEE-DID model after IPTW-weighting. Table 1 Baseline characteristics of study cohort

Characteristics	Controls	Acupuncture	 P value	
N (%)	16129 (97.0%)	495 (3.0%)	
Year of index date, n (%)			0.010*	
 2007	1368 (8.5%)	58 (11.7%)	
 2008	1903 (11.8%)	48 (9.7%)	
 2009	1623 (10.1%)	57 (11.5%)	
 2010	1504 (9.3%)	28 (5.7%)	
 2011	1360 (8.4%)	39 (7.9%)	
 2012	1309 (8.1%)	48 (9.7%)	
 2013	1067 (6.6%)	32 (6.5%)	
 2014	984 (6.1%)	25 (5.1%)	
 2015	1145 (7.1%)	42 (8.5%)	
 2016	1135 (7.0%)	40 (8.1%)	
 2017	840 (5.2%)	21 (4.2%)	
 2018	724 (4.5%)	19 (3.8%)	
 2019	580 (3.6%)	27 (5.5%)	
 2020	567 (3.5%)	11 (2.2%)	
 2021	20 (0.1%)	0 (0.0%)	
Index age, median (Q1, Q3)	55 (50, 60)	55 (50, 59)	 0.83	
Age at breast cancer diagnosis, median (Q1, Q3)	53 (47, 57)	52 (47, 57)	 0.11	
Female, n (%)	15,977 (99.1%)	492 (99.4%)	 0.73	
US region, n (%)			< 0.001***	
 Northeast	3297 (20.4%)	88 (17.8%)	
 Midwest	4760 (29.5%)	61 (12.3%)	
 South	3996 (24.8%)	22 (4.4%)	
 West	3669 (22.7%)	323 (65.3%)	
Payer type, n (%)			< 0001***	
 Commercial	13,927 (86.3%)	455 (91.9%)	
 Managed Medicaid	481 (3.0%)	7 (1.4%)	
 Medicare advantage or Medicare cost	458 (2.8%)	3 (0.6%)	
 Others	1263 (7.8%)	30 (6.1%)	
Plan type, n (%)			< 0.001***	
 PPO	9683 (60.0%)	366 (73.9%)	
 HMO	3411 (21.1%)	55 (11.1%)	
 POS	770 (4.8%)	22 (4.4%)	
 Others	2265 (14.0%)	52 (10.5%)	
Time from pre-index pain health encounter (in days), median (Q1, Q3)	252 (119, 336)	236 (105, 328)	 0.14	
Pain subtypes, n (%) 	
 Musculoskeletal	15,642 (97.0%)	486 (98.2%)	 0.12	
 General	1188 (7.4%)	46 (9.3%)	 0.11	
 Neoplasm-related	240 (1.5%)	5 (1.0%)	 0.38	
Annual healthcare cost (2022 USD), median (Q1, Q3)	
 All claims	12,779.45 (5735.13, 34,230.06)	21,492.51 (9591.98, 62,439.71)	< 0.001***	
 Pain-related claims	457.58 (187.25, 1390.70)	902.03 (435.49, 2251.74)	< 0.001***	
NCI Charlson’s comorbidity index, n (%)			0.28	
 0	10,439 (64.7%)	320 (64.6%)	
 1	3761 (23.3%)	128 (25.9%)	
 2	1227 (7.6%)	31 (6.3%)	
 3 or more	702 (4.4%)	16 (3.2%)	
Cancer-related characteristics, n (%)	
 Presence of metastases	1244 (7.7%)	57 (11.5%)	 0.002**	
 Bone metastases	601 (3.7%)	21 (4.2%)	 0.55	
 Brain metastases	133 (0.8%)	3 (0.6%)	 0.59	
 Prior taxane exposure	3038 (18.8%)	122 (24.6%)	 0.001**	
 Current or prior tamoxifen exposure	2143 (13.3%)	65 (13.1%)	 0.92	
 Current or prior aromatase inhibitors exposure	2273 (14.1%)	75 (15.2%)	 0.51	
Supportive care-related characteristics, n (%)	
 Current antidepressant use	3954 (24.5%)	131 (26.5%)	 0.32	
 Depression	2739 (17.0%)	91 (18.4%)	 0.41	
 Two or more 30-day supply of opioid prescription fills	659 (4.1%)	14 (2.8%)	 0.16	
 Menopausal symptoms	2091 (13.0%)	133 (26.9%)	< 0.001***	
Abbreviations: HMO Health maintenance organization, N/n, counts, NCI National Cancer Institute, POS Point-of-service, PPO Preferred provider organization, Q1 Quartile 1, Q3, Quartile 3; USD United States Dollar

*P< 0.05, **P< 0.01, ***P< 0.001

All analyses were two-tailed, tested at 5% significance level, and presented as ratios or percentage changes accompanying 95% confidence intervals (CI).

Sensitivity analysis

To emulate a randomized controlled trial design, sensitivity analysis comparing acupuncture-treated and non-treated cohorts utilized a 1:1 nearest neighbor matching without replacement. Propensity scores were estimated via logistic regression, adjusting for all covariates listed in Table 1.

Exploratory analysis

Because acupuncture treatment is highly heterogenous in terms of treated conditions [34, 35] and real-world utilization rate [36], we conducted exploratory analyses comparing changes in outcomes between (1) low (< 10 sessions) against high (≥ 10 sessions [18]) acupuncture utilization and (2) acupuncture treatment for pain against treatment for other conditions.

Data extraction, cleaning, IPTW, and propensity score matching (PSM, with R package MatchIt [37]) were executed with R version 4.3.2 [38]. All other analyses were performed on Stata version 16.1.

Results

Cohort characteristics

Our cohort comprised 16,624 eligible breast cancer survivors (median age = 55 years; stage 4 = 7.8%) who averaged 2.5 years (standard deviation, SD = 1.9) after cancer diagnosis. Four hundred ninety-five (3.0%) survivors had received acupuncture within a year of pain-related health encounter. Compared to non-treated survivors, more acupuncture-treated survivors had metastatic cancer, prior taxane chemotherapy exposure, reported more menopausal symptoms, lived in the Western region of the US, and with commercial PPO health plans. Acupuncture-treated patients were also with higher healthcare expenditure during the pre-index period (P < 0.05, Table 1). The selection algorithm for eligible patients is illustrated in Additional file 1: Fig. S1. Cost outcomes, both pre- and post-index, were found with a right-skewed distribution (Additional file 1: Fig. S2).

Acupuncture utilization

Acupuncture-treated patients completed an average of 11.5 (median = 6) acupuncture sessions in a given year. The total annual cost of acupuncture per patient was $919.60 (median = $476.81), paying $183.22 (median = $49.12) out-of-pocket (OOP) on average. Cost per session averaged $84.80 (median = $78.56), with $22.39 (median = $10.17) being OOP expenditure (Additional file 1: Table S2).

Impact of acupuncture on analgesics utilization

Opioid utilization

After acupuncture initiation, the prevalence of opioid users reduced by 36% (95% CI = 0.53 to 0.85, P < 0.01). The DID was estimated to have a prevalence ratio (PR) of 0.80 (95% CI = 0.66 to 0.96, P < 0.05), which indicated that the relative prevalence of opioid use decreased by 20% in the acupuncture-treated group compared to non-treated patients, over the 2-year, pre-index to post-index periods (Table 2, Fig. 2A). However, we did not observe statistically significant DID estimate on opioid prescription fill rates (Table 2, Fig. 2B–C). All predetermined covariates were significantly associated with opioid use in the bivariate analyses, except for biological sex, current or previous tamoxifen exposure, menopausal symptoms, and having two or more 30-day opioid prescription fills (Additional file 1: Table S3). Table 2 Difference-in-difference analysis comparing changes in outcomes among acupuncture against controls

	Acupuncture (N = 495)	Controls, ref (N = 16129)	DID	Controls PSM, ref (N = 495)	DID	
	Pre-index	 Post-index	  Pre-post change ratioa (95% CI)	  Pre-index	  Post-index	  Pre-post change ratiob (95% CI)	  Ratioc (95% CI)	  IPTW-ratiod (95% CI)	  Pre-index	  Post-index	 Pre-post change ratiob (95% CI)	 PSM-ratioe (95% CI)	
Proportion of users, n (%)	
 Opioids	145 (29.3%)	93 (18.8%)	0.64*** (0.53, 0.77)	4191 (26.0%)	3381 (21.0%)	0.81*** (0.78, 0.83)	0.80* (0.66, 0.96)	0.88 (0.68, 1.15)	136 (27.5%)	99 (20.0%)	0.73** (0.61, 0.87)	0.88 (0.68, 1.14)	
 NSAIDs	103 (20.8%)	70 (14.1%)	0.68** (0.54, 0.85)	3100 (19.2%)	2605 (16.2%)	0.84*** (0.81, 0.87)	0.81 (0.65, 1.01)	0.87 (0.56, 1.34)	86 (17.4%)	66 (13.3%)	0.77* (0.60, 0.98)	0.89 (0.64, 1.23)	
 SNRIs	47 (9.5%)	48 (9.7%)	1.02 (0.79, 1.33)	1402 (8.7%)	1387 (8.6%)	0.99 (0.94, 1.04)	1.04 (0.79, 1.35)	0.80 (0.55, 1.17)	66 (13.3%)	56 (11.3%)	0.83 (0.66, 1.04)	1.23 (0.87, 1.75)	
 TCAs	18 (3.6%)	12 (2.4%)	0.66 (0.39, 1.10)	433 (2.7%)	423 (2.6%)	0.98 (0.89, 1.07)	0.67 (0.40, 1.14)	0.39 (0.12, 1.24)	13 (2.6%)	14 (2.8%)	1.08 (0.61, 1.92)	0.61 (0.28, 1.32)	
 Gabapentinoids	47 (9.5%)	65 (13.1%)	1.44** (1.10, 1.89)	1272 (7.9%)	1228 (7.6%)	0.96 (0.91, 1.02)	1.50** (1.14, 1.97)	1.53* (1.06, 2.23)	41 (8.3%)	41 (8.3%)	1.00 (0.73, 1.37)	1.44 (0.95, 2.18)	
Short-term Rx fills, count (rate/person)	
 Opioids	263 (0.53)	243 (0.49)	0.92 (0.68, 1.26)	9866 (0.61)	7922 (0.49)	0.80*** (0.77, 0.84)	1.15 (0.84, 1.58)	0.84 (0.51, 1.38)	329 (0.66)	209 (0.42)	0.64** (0.48, 0.84)	1.45 (0.96, 2.20)	
 NSAIDs	105 (0.21)	77 (0.16)	0.73 (0.52, 1.04)	2202 (0.14)	1706 (0.11)	0.77*** (0.71, 0.84)	0.95 (0.66, 1.36)	0.93 (0.59, 1.47)	68 (0.14)	60 (0.12)	0.88 (0.57, 1.38)	0.83 (0.47, 1.46)	
 SNRIs	9 (0.02)	10 (0.02)	1.11 (0.40, 3.06)	202 (0.01)	168 (0.01)	0.83 (0.60, 1.16)	1.34 (0.46, 3.88)	0.81 (0.29, 2.23)	10 (0.02)	2 (0.00)	0.20* (0.05, 0.82)	5.56 (0.98, 31.62)	
 TCAs	1 (0.00)	1 (0.00)	1.00 (0.06, 15.99)	85 (0.01)	73 (0.00)	0.86 (0.54, 1.35)	1.16 (0.07, 19.32)	1.43 (0.09, 23.79)	3 (0.01)	1 (0.00)	0.33 (0.02, 5.34)	3.00 (0.06, 151.49)	
 Gabapentinoids	10 (0.02)	4 (0.02)	0.40 (0.10, 1.58)	325 (0.02)	287 (0.01)	0.88 (0.67, 1.16)	0.45 (0.11, 1.83)	4.19 (0.61, 28.80)	21 (0.04)	17 (0.03)	0.81 (0.42, 1.56)	0.49 (0.11, 2.26)	
Long-term Rx fills, count (rate/person)	
 Opioids	82 (0.17)	109 (0.22)	1.33 (0.81, 2.17)	4694 (0.29)	5180 (0.32)	1.10** (1.04, 1.17)	1.20 (0.73, 1.98)	0.57 (0.15, 2.18)	85 (0.17)	101 (0.20)	1.19 (0.82, 1.72)	1.12 (0.60, 2.07)	
 NSAIDs	153 (0.31)	177 (0.36)	1.16 (0.80, 1.68)	5046 (0.31)	5022 (0.31)	1.00 (0.94, 1.05)	1.16 (0.80, 1.69)	1.32 (0.72, 2.41)	143 (0.29)	122 (0.25)	0.85 (0.61, 1.19)	1.36 (0.82, 2.24)	
 SNRIs	198 (0.40)	217 (0.44)	1.10 (0.85, 1.41)	7098 (0.44)	7170 (0.44)	1.01 (0.96, 1.06)	1.08 (0.84, 1.40)	0.97 (0.76, 1.25)	324 (0.65)	323 (0.65)	1.00 (0.80, 1.24)	1.10 (0.79, 1.54)	
 TCAs	70 (0.14)	47 (0.09)	0.67 (0.36, 1.25)	1677 (0.10)	1781 (0.11)	1.06 (0.96, 1.17)	0.63 (0.34, 1.19)	0.33 (0.10, 1.10)	67 (0.14)	52 (0.11)	0.78 (0.46, 1.30)	0.87 (0.38, 1.94)	
 Gabapentinoids	186 (0.38)	295 (0.60)	1.59** (1.21, 2.08)	4451 (0.28)	4841 (0.30)	1.09* (1.02, 1.16)	1.46** (1.10, 1.93)	1.21 (0.94, 1.55)	145 (0.29)	204 (0.41)	1.41 (0.70, 2.82)	1.13 (0.53, 2.38)	
Healthcare resource utilization	
 Total cost (2022 USD), median (Q1, Q3)	21,493 (9592, 62,440)	16,027 (7878, 33,534)	0.74** (0.61, 0.89)	12,779 (5735, 34,230)	9183 (4010, 21,805)	0.72*** (0.69, 0.74)	1.03 (0.85, 1.24)	1.29* (1.09, 1.52)	19,242 (9438, 58,526)	11,126 (4784, 26,870)	0.59*** (0.49, 0.71)	1.25 (0.96, 1.62)	
 Hospitalization, count (rate/person)	105 (0.21)	82 (0.17)	0.78 (0.56, 1.10)	2906 (0.18)	2232 (0.14)	0.77*** (0.72, 0.82)	1.02 (0.72, 1.44)	1.72* (1.04, 2.86)	122 (0.25)	71 (0.14)	0.58** (0.43, 0.79)	1.34 (0.85, 2.13)	
 Emergency department, count (rate/person)	206 (0.42)	175 (0.35)	0.85 (0.65, 1.11)	7307 (0.45)	5939 (0.37)	0.81*** (0.78, 0.85)	1.05 (0.80, 1.37)	1.29 (0.92, 1.80)	220 (0.44)	131 (0.26)	0.60*** (0.47, 0.76)	1.43 (0.99, 20.5)	
Abbreviations: CI, confidence interval; DID, difference-in-difference; GEE, generalized estimating equation; IPTW, inverse probability of treatment weighting; N/n, counts; NSAIDs, non-steroidal anti-inflammatory drugs; PSM, propensity score matched; Q1, quartile 1; Q3, quartile 3; ref, reference group for DID analysis; Rx, prescription; SNRIs, serotonin-norepinephrine reuptake inhibitors; TCAs, tricyclic antidepressants; USD, United States dollar

*P < 0.05, **P < 0.01, ***P < 0.001

aExponential function of β1 + β3 in the GEE-DID model, log(Y) = β0 + β1*[post-index] + β2*[acupuncture group] + β3*[post-index × acupuncture group] + ε, where Y are either the prevalence or odds for binary outcomes, rate outcomes for prescription fills, hospitalization and emergency department visits, or cost outcomes

bExponential function of β1 in the GEE-DID model

cExponential function of β3 in the GEE-DID model

dExponential function of β3 in the GEE-DID model after IPTW-weighting. Time from pre-index pain health encounter could not achieve a standardized mean difference <|0.1| after IPTW-weighting and was thus added into the models (eFig. 3)

eExponential function of β3 in the GEE-DID model after PSM

Fig. 2 Percentage changes with 95% confidence intervals (pre-index to post-index) in opioid and NSAID utilization outcomes. The percentage change (pre-post) statistics, with 95% confidence intervals, were calculated from the pre-post change ratios from Tables 2 and 3. For example, a pre-post change ratio of 0.64 for opioid users among acupuncture-exposed patients (n = 495, Table 2), the percentage change (pre-post) statistics will be (0.64 − 1) × 100% =  − 36%. Absolute pre-index and post-index outcomes for the subgroups are presented in Table 2, eTable 6, and eTable 8. *P < 0.05 (DID estimate)

The characteristics and covariate balance statistics of the IPTW-weighted and PSM-matched cohorts are summarized in eFig. 3 and Additional file 1: Table S4, respectively. After IPTW and PSM, the DID estimates on opioid use prevalence was not significantly different between treated and non-treated patients (Table 2, Fig. 2A–C).

NSAID utilization

Acupuncture treatment was associated with a 32% decline (95% CI = 0.54 to 0.85, P < 0.01) in the prevalence of NSAID users from baseline, with a DID PR estimate of 0.81, albeit not statistically significant (95% CI = 0.65 to 1.01, P > 0.05), compared to non-treated patients (Table 2, Fig. 2D). We did not observe any significant DID estimate on NSAID prescription fill rates or after IPTW and PSM (Table 2, Fig. 2E–F).

Adjuvant analgesics utilization

Acupuncture was not associated with any changes in TCA or SNRI utilization. However, significant increases in prevalence (odds ratio [OR] = 1.44, 95% CI = 1.10 to 1.89, P < 0.01) and long-term prescription fills (rate ratio [RR] = 1.59, 95% CI = 1.21 to 2.08, P < 0.01) of gabapentinoids were found among acupuncture-treated patients. This effect was associated with DID ratio estimate of 1.50 and 1.46, respectively, when compared to non-treated patients (P < 0.01). All predetermined covariates were significantly associated with gabapentinoid use, except biological sex and tamoxifen exposure. Age at breast cancer diagnosis was also significantly associated with rates of long-term prescription fills of gabapentinoids (Additional file 1: Table S3). DID remained statistically significant in the IPTW-weighted model but not after PSM when evaluating odds of gabapentinoid use (Table 2).

Impact of acupuncture on healthcare cost and resource utilization

Breast cancer survivors who initiated acupuncture were found with an average decline in total healthcare cost by 26% (95% CI = 0.61 to 0.89, P < 0.01) and an absolute decrease by $15,513 (95% CI =  − $24,815 to − $6213, P < 0.01). This decrease was not significantly different from non-treated patients. However, DID estimates differ between IPTW and PSM analyses (Table 2).

Exploratory analysis 1: high vs low acupuncture utilization

One hundred seventy-eight (36.0%) patients received ≥ 10 sessions and were characterized by a higher overall healthcare expenditure and increased prior taxane exposure at baseline (Additional file 1: Table S5).

The annualized changes in analgesics and healthcare resource utilization outcomes were not significantly different between low (< 10 sessions) and high (≥ 10 sessions) acupuncture utilization breast cancer survivors. Notably, a significantly larger increase in gabapentinoid utilization was observed among low-utilization patients (P < 0.01) but not among those with higher acupuncture utilization (Table 3, Additional file 1: Table S6). Table 3 Difference-in-difference exploratory analysis by (1) high vs low acupuncture utilization and (2) acupuncture for pain vs other conditions

	Exploratory analysis 1: high vs low acupuncture utilization	Exploratory analysis 2: acupuncture for pain vs other conditions	
	  Acupuncture, ≥ 10 sessions (N = 178)	  Acupuncture, < 10 sessions (N = 317)	  DID (ref: < 10 sessions)	  Acupuncture for pain (N = 264)	  Acupuncture for other conditions (N = 231)	  DID (ref: other conditions)	
	  Pre-post change ratioa (95% CI)	  Pre-post change ratiob (95% CI)	  Ratioc (95% CI)	  Pre-post change ratioa (95% CI)	  Pre-post change ratiob (95% CI)	  Ratioc (95% CI)	
Proportion of users, PR/OR	
 Opioids	0.59** (0.43, 0.80)	0.67** (0.53, 0.85)	0.88 (0.59, 1.30)	0.53*** (0.40, 0.70)	0.77* (0.60, 0.99)	0.69 (0.47, 1.00)	
 NSAIDs	0.67* (0.48, 0.95)	0.68* (0.51, 0.91)	0.98 (0.63, 1.55)	0.63** (0.47, 0.85)	0.74 (0.53, 1.05)	0.85 (0.54, 1.34)	
 SNRIs	0.79 (0.49, 1.26)	1.14 (0.83, 1.57)	0.69 (0.39, 1.22)	1.11 (0.74, 1.66)	0.96 (0.68, 1.35)	1.16 (0.68, 1.97)	
 TCAs	0.74 (0.36, 1.52)	0.59 (0.28, 1.23)	1.25 (0.45, 3.50)	1.00 (0.56, 1.78)	0.24* (0.07, 0.82)	4.11* (1.07, 15.72)	
 Gabapentinoids	1.17 (0.79, 1.72)	1.66** (1.15, 2.41)	0.70 (0.41, 1.20)	1.86** (1.26, 2.75)	1.13 (0.77, 1.64)	1.65 (0.96, 2.84)	
Short-term Rx fills, RR	
 Opioids	0.97 (0.53, 1.75)	0.90 (0.63, 1.28)	1.08 (0.54, 2.14)	0.65* (0.45, 0.93)	1.24 (0.78, 1.95)	0.53* (0.29, 0.94)	
 NSAIDs	0.52* (0.30, 0.91)	0.87 (0.58, 1.32)	0.60 (0.30, 1.20)	0.58* (0.37, 0.91)	1.14 (0.70, 1.86)	0.51* (0.26, 0.99)	
 SNRIs	0.50 (0.03, 8.00)	1.29 (0.42, 3.91)	0.39 (0.02, 7.73)	0.33 (0.03, 3.85)	1.50 (0.44, 5.14)	0.22 (0.01, 3.44)	
 TCAsd	–	–	–	–	–	–	
 Gabapentinoidsd	–	–	–	0.20 (0.02, 2.54)	0.60 (0.13, 2.79)	0.33 (0.02, 6.50)	
Long-term Rx fills, RR	
 Opioids	1.19 (0.33, 4.25)	1.38 (0.83, 2.30)	0.86 (0.22, 3.41)	1.53 (0.88, 2.68)	1.10 (0.48, 2.55)	1.39 (0.51, 3.81)	
 NSAIDs	0.97 (0.57, 1.63)	1.28 (0.79, 2.08)	0.76 (0.37, 1.54)	1.03 (0.66, 1.62)	1.32 (0.73, 2.38)	0.78 (0.37, 1.64)	
 SNRIs	0.84 (0.57, 1.23)	1.20 (0.88, 1.63)	0.70 (0.43, 1.14)	1.37 (0.86, 2.17)	0.93 (0.72, 1.19)	1.48 (0.87, 2.50)	
 TCAs	0.44 (0.18, 1.08)	0.91 (0.41, 2.03)	0.49 (0.15, 1.61)	0.78 (0.39, 1.55)	0.27 (0.06, 1.23)	2.93 (0.55, 15.70)	
 Gabapentinoids	1.31 (0.86, 2.01)	1.88*** (1.35, 2.61)	0.70 (0.41, 1.20)	1.99** (1.19, 3.31)	1.33 (0.99, 1.80)	1.49 (0.82, 2.69)	
Healthcare resource utilization	
 Total cost (2022 USD), cost change ratio	0.59*** (0.46, 0.76)	0.82 (0.64, 1.06)	0.72 (0.50, 1.02)	0.67*** (0.54, 0.82)	0.79 (0.59, 1.06)	0.85 (0.59, 1.21)	
 Hospitalization, RR	0.56 (0.30, 1.04)	0.92 (0.61, 1.38)	0.61 (0.29, 1.28)	0.72 (0.41, 1.26)	0.83 (0.54, 1.27)	0.86 (0.43, 1.75)	
 Emergency department, RR	0.77 (0.52, 1.14)	0.93 (0.64, 1.33)	0.83 (0.48, 1.41)	0.84 (0.57, 1.23)	0.86 (0.59, 1.25)	0.97 (0.57, 1.66)	
Abbreviations: CI, confidence interval; DID, difference-in-difference; GEE, generalized estimating equation; N/n, counts; NSAIDs, non-steroidal anti-inflammatory drugs; OR, odds ratio; PR; prevalence ratio; ref, reference group for DID analysis; RR, rate ratio; Rx, prescription; SNRIs, serotonin-norepinephrine reuptake inhibitors; TCAs, tricyclic antidepressants; USD, United States dollar

*P < 0.05, **P < 0.01, ***P < 0.001

aExponential function of β1 + β3 in the GEE-DID model, log(Y) = β0 + β1*[post-index] + β2*[group] + β3*[post-index × group] + ε, where Y are either the prevalence or odds for binary outcomes, rate outcomes for prescription fills, hospitalization and emergency department visits, or cost outcomes

bExponential function of β1 in the GEE-DID model

cExponential function of β3 in the GEE-DID model

dModels did not converge if point estimate and 95% CI is not estimated

High-utilization patients, despite spending more on acupuncture sessions (average total cost: $1035 vs $787, P = 0.054), were observed with a significant reduction in total healthcare costs after acupuncture treatment (P < 0.001) unlike low-utilization patients (Table 3, Additional file 1: Table S6). After further stratifying into quintiles based on number of completed acupuncture sessions, the average cost savings in quintiles 1, 2, 4, and 5 were comparable to the PSM-control cohort, except for quintile 3 which comprised patients who completed 5 to 8.5 sessions (Additional file 1: Fig. S4).

Exploratory analysis 2: acupuncture for pain vs other conditions

Two hundred sixty-four (53.3%) patients received acupuncture specifically for pain and, compared to patients treated for other conditions (n = 231), were observed with less neoplasm-related pain diagnoses and more acupuncture sessions (Additional file 1: Table S7).

Patients receiving acupuncture for pain were found with statistically significant DID RR estimates of 0.53 (95% CI = 0.29 to 0.94, P < 0.05) and 0.51 (95% CI = 0.26 to 0.99, P < 0.05), suggesting a decline in short-term opioid and NSAID prescription fills compared to patients using acupuncture for treating other conditions (Table 3, Fig. 2B and F). These DID estimates remained statistically significant after adjusting for neoplasm-related pain diagnoses (P < 0.05, results not presented). On the other hand, patients treated for non-pain conditions were observed with a significantly larger decline in prevalence of TCA users compared to those treated for pain (P < 0.05, Table 3, Additional file 1: Table S8). We also observed a statistically significant increase in gabapentinoid utilization (P < 0.01) and reduction in total healthcare cost (P < 0.001) only among patients treated for pain (Table 3, Additional file 1: Table S8). Further stratified analysis found that gabapentinoid utilization was significantly increased among low-utilization pain-treated survivors (P < 0.05) rather than high-utilization patients (P > 0.05) (Additional file 1: Table S9).

 Discussion

This is the first study that has evaluated the real-world impact of acupuncture treatment for cancer pain. The overall acupuncture utilization for pain is low (< 3%) among commercially insured breast cancer survivors, albeit comparable to previous published data [39]. Across all analyzed subgroups of patients, those treated specifically for pain were found with the largest decline in prevalence and short-term fills of opioids and NSAIDs after acupuncture treatment. High-utilization patients were found with greater healthcare cost savings compared to low-utilization patients, with the latter escalating to the use of gabapentinoids. Taken together, findings confirm that acupuncture reduces pain and that a value-based care model may be most appropriate to facilitate increased acupuncture coverage for pain management among cancer survivors.

Acupuncture analgesia has been widely recognized in multiple clinical guidelines in both cancer and non-cancer patient populations [12, 40, 41] and functionally ascribed to local and systemic neurochemical changes. The role of endogenous opiates in acupuncture analgesia was evident from the lack of analgesic response after naloxone administration and opiate receptors depletion [42, 43]. Local release of adenosines and the activation of adenosine A1 receptors, which serve as inhibitory mediators for neurons, are crucial in inhibiting the transmission of painful stimuli to the anterior cingulate cortex during acupuncture [44]. Clinical evidence for acupuncture in cancer pain is aplenty, with two large, randomized trials specifically described in the SIO-ASCO integrative oncology guidelines for pain management [12]. The PEACE study (n = 360) investigated the efficacy of electroacupuncture and auricular acupuncture compared to waitlist control reported significant pain reduction after 10 weekly acupuncture sessions [18]. Another trial recruited breast cancer patients (n = 226) experiencing aromatase inhibitor-induced joint pain and found that 12 weeks (18 sessions) of acupuncture had relieved pain compared to sham and waitlist controls [17]. Our study contributed by finding a decline in opioid and NSAID real-world medication use between the year before and after acupuncture initiation. Importantly, this improvement is not attributable to non-specific effects of acupuncture as evidenced by the lack of effectiveness observed among patients treated for non-pain conditions. In all, acupuncture is not just a placebo as it has achieved a clinically meaningful impact on patients’ health in the real-world.

Some pain-treated, low-utilization patients were observed with increased gabapentinoid usage. Considering the relative safety of acupuncture across indications [45], we hypothesize that gabapentinoid initiation is confined to non-responders of acupuncture. Relatedly, our assessment of total healthcare expenditure concluded that cost savings after acupuncture initiation were comparable to non-treated controls, other than one-fifth of treated patients who were not frequent users of acupuncture and likely non-responders. Thus, future research should examine ways to predict acupuncture response prior to treatment initiation. Applying the concept of precision medicine, the research into clinical, sociodemographic, environmental, lifestyle, and genetic predictors of treatment response will enable the prescription of acupuncture to patients who will receive maximal health benefits [46]. Adopting a value-based care model by periodically evaluating patient response to acupuncture (e.g., monitoring patient-reported outcomes and ensuring patient adherence) will ensure that only responders should continue with the treatment and prevent unnecessary healthcare expenditure [47]. Furthermore, it is known that acupuncture utilization was more frequent among individuals of female gender, White or Asian racial-ethnic background, and those with higher levels of education and income due to affordability-related barriers [21, 48, 49]. Despite the promises of acupuncture for cancer pain management, the inequitable access to acupuncture for minoritized and marginalized populations will pervade the widening care disparities [21, 50]. Other payers, including Medicaid, Medicare, and HMOs (referring to Additional file 1: Table S3, these payers were observed with higher prescription rates of analgesics), should consider implementing precision medicine and value-based care to enhance cancer survivors’ access to acupuncture treatment for pain and reduce healthcare disparities.

There are several limitations in this study. As with other retrospective studies, unmeasured confounding is unavoidable. Although acupuncture is recommended for all cancer survivors, we have focused only breast cancer survivors to minimize the degree of unmeasured confounding; future studies should investigate in another group of cancer survivors. Even so, not all sociodemographic variables are available, thus we could not adjust for factors such as race/ethnicity and education attainment that are different between acupuncture-treated and non-treated patients. Nevertheless, while these characteristics may impact acupuncture access and utilization, they are unlikely to impact the underlying efficacy [50]. Selection bias is another concern. Naturally, cancer patients seeking integrative therapies such as acupuncture have greater symptom burden and poorer symptom control than their peers [51–53]. While we have attempted to control for potential confounders such as antidepressants and chronic opioid use, our PSM-control cohort is observed with greater improvements in ED and hospitalization rates compared to other control and acupuncture-treated patients. Non-pharmacological modalities for pain relief are seldom captured in claims dataset, in vie that they are often out-of-pocket expenditures [39]. Patients seeking these modalities are likely to share similar characteristics to the patients within the acupuncture-treated cohort in our study, thus possibly explaining the lack of associations observed in our analyses after IPTW and PSM. Acupuncture utilization was also highly heterogenous, with only 53% receiving it for pain management. We reason that these limitations have blunted the reported association, and the true effect size may be greater.

Finally, our statistical power and study design are impacted by the low real-world utilization of reimbursed acupuncture treatment. In Medicare Part B, beginning in 2020, acupuncture is claimable for only chronic low back pain [54]. Claims for other pain-related indications can only be obtained from Medicare managed care (i.e., Medicare Part C). The consequence is that SEER-Medicare, the database of choice for cancer-specific questions among older adults, is not suitable as the reliability and completeness of Medicare Part C data remains an area of active research [55, 56]. It is prudent to apply this methodology on SEER-Medicare after data validations to address the effectiveness of acupuncture coverage in older cancer survivors. Regardless, the study’s strength is underscored by its large sample size of over 10,000 patients and a robust pre-post-intervention, DID study design that ensures exposure-outcome temporality and similar index times for comparison of outcomes between acupuncture and control patients, effectively answering an important research question regarding the real-world treatment effect of acupuncture at managing pain in breast cancer survivors.

Conclusion

Although adjusted results did not show that patients receiving acupuncture had better outcomes than non-treated patients, exploratory analyses revealed that patients treated specifically for pain used fewer analgesics and those with high acupuncture utilization incurred lower healthcare costs. Follow-up research should evaluate ways to examine acupuncture effectiveness in real-world settings with pragmatic trials, understand predictors of acupuncture response, validate our findings within SEER-Medicare, and replicate with other groups of cancer survivors (e.g., bone metastases). Clinicians and payers can consider applying value-based care to enhance cancer survivors’ access to acupuncture treatment for pain.

Supplementary Information

Supplementary Material 1: Additional file 1: Table S1 Algorithms for identifying health-related covariates. Table S2 Acupuncture utilization statistics. Table S3 Bivariate association analysis between predetermined covariates and outcomes with significant DID estimates when comparing acupuncture-treated vs non-treated breast cancer survivors. Table S4 Baseline characteristics comparing acupuncture against control patients after propensity score matching. Table S5 Baseline characteristics of high vs low acupuncture utilization patients. Table S6 Difference-in-difference exploratory analysis by high vs low acupuncture utilization. Table S7 Baseline characteristics comparing acupuncture-treated patients for pain vs for other conditions. Table S8 Difference-in-difference exploratory analysis by acupuncture for pain vs other conditions. Table S9 Change in gabapentinoid utilization and total healthcare cost among patients treated with acupuncture for pain, stratified by high vs low acupuncture utilization. Fig. S1 Acupuncture and control cohort selection. Fig. S2 Distribution of annualized total healthcare cost, pre- and post-index. Fig. S3 Standardized mean differences in the unadjusted (original) and weighted (IPTW-weighted) cohorts. Fig. S4 Effect of number of acupuncture sessions (in quintiles) on the change in total all-cause healthcare cost from pre- to post-index.

Acknowledgements

We acknowledge Chapman University School of Pharmacy for acquiring the data and IQVIA for reviewing this work. We acknowledge that any opinions, findings, conclusions, or recommendations expressed in this publication are those of the author(s) and do not necessarily reflect the views of IQVIA. We thank Drs Farhad Islami, Chia Jie Tan, and Daniel Wiese for their expert advice during the planning and analysis phases of the study. The authors also wish to acknowledge the support of the Chao Family Comprehensive Cancer Center Biostatistics Shared Resource, supported by the National Cancer Institute of the National Institutes of Health under award number P30CA062203. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Authors’ contributions

All authors read and approved the final manuscript. DQN: conception and design of work; analysis and interpretation of data; drafted and substantially revised manuscript. SL: interpretation of data. RTL: interpretation of data. YW: acquisition and interpretation of data; substantially revised manuscript. AC: interpretation of data; substantially revised manuscript.

Funding

This study was supported by the PhRMA Foundation Predoctoral Fellowship in Value Assessment-Health Outcomes (to Mr. Ng). The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Availability of data and materials

The dataset was available for purchase from IQVIA. We do not own the data and hence are not permitted to share in the original form. The underlying code for this study is not publicly available but may be made available to qualified researchers on reasonable request from the corresponding authors.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
==== Refs
References

1. Miller KD Nogueira L Mariotto AB Cancer treatment and survivorship statistics, 2019 CA Cancer J Clin 2019 69 5 363 385 10.3322/caac.21565 31184787
Miller KD, Nogueira L, Mariotto AB, et al. Cancer treatment and survivorship statistics, 2019. CA Cancer J Clin. 2019;69(5):363–85.31184787
2. Hamer J McDonald R Zhang L Quality of life (QOL) and symptom burden (SB) in patients with breast cancer Support Care Cancer 2017 25 2 409 419 10.1007/s00520-016-3417-6 27696078
Hamer J, McDonald R, Zhang L, et al. Quality of life (QOL) and symptom burden (SB) in patients with breast cancer. Support Care Cancer. 2017;25(2):409–19.27696078
3. Morrison EJ Novotny PJ Sloan JA Emotional problems, quality of life, and symptom burden in patients with lung cancer Clin Lung Cancer 2017 18 5 497 503 10.1016/j.cllc.2017.02.008 28412094
Morrison EJ, Novotny PJ, Sloan JA, et al. Emotional problems, quality of life, and symptom burden in patients with lung cancer. Clin Lung Cancer. 2017;18(5):497–503.28412094
4. van den Beuken-van Everdingen MH Hochstenbach LM Joosten EA Tjan-Heijnen VC Janssen DJ Update on prevalence of pain in patients with cancer: systematic review and meta-analysis J Pain Symptom Manage 2016 51 6 1070 1090 e1079 10.1016/j.jpainsymman.2015.12.340 27112310
van den Beuken-van Everdingen MH, Hochstenbach LM, Joosten EA, Tjan-Heijnen VC, Janssen DJ. Update on prevalence of pain in patients with cancer: systematic review and meta-analysis. J Pain Symptom Manage. 2016;51(6):1070-1090 e1079.27112310
5. Haenen V Evenepoel M De Baerdemaecker T Pain prevalence and characteristics in survivors of solid cancers: a systematic review and meta-analysis Support Care Cancer 2022 31 1 85 10.1007/s00520-022-07491-8 36574040
Haenen V, Evenepoel M, De Baerdemaecker T, et al. Pain prevalence and characteristics in survivors of solid cancers: a systematic review and meta-analysis. Support Care Cancer. 2022;31(1):85.36574040
6. Ventafridda V Saita L Ripamonti C De Conno F WHO guidelines for the use of analgesics in cancer pain Int J Tissue React 1985 7 1 93 96 2409039
Ventafridda V, Saita L, Ripamonti C, De Conno F. WHO guidelines for the use of analgesics in cancer pain. Int J Tissue React. 1985;7(1):93–6.2409039
7. Heer E Harper A Escandor N Sung H McCormack V Fidler-Benaoudia MM Global burden and trends in premenopausal and postmenopausal breast cancer: a population-based study Lancet Glob Health 2020 8 8 e1027 e1037 10.1016/S2214-109X(20)30215-1 32710860
Heer E, Harper A, Escandor N, Sung H, McCormack V, Fidler-Benaoudia MM. Global burden and trends in premenopausal and postmenopausal breast cancer: a population-based study. Lancet Glob Health. 2020;8(8):e1027–37.32710860
8. Acheampong T Kehm RD Terry MB Argov EL Tehranifar P Incidence trends of breast cancer molecular subtypes by age and race/ethnicity in the US from 2010 to 2016 JAMA Netw Open 2020 3 8 e2013226 10.1001/jamanetworkopen.2020.13226 32804214
Acheampong T, Kehm RD, Terry MB, Argov EL, Tehranifar P. Incidence trends of breast cancer molecular subtypes by age and race/ethnicity in the US from 2010 to 2016. JAMA Netw Open. 2020;3(8): e2013226.32804214
9. Grigorian N Baumrucker SJ Aromatase inhibitor-associated musculoskeletal pain: an overview of pathophysiology and treatment modalities SAGE Open Med 2022 10 20503121221078722 10.1177/20503121221078722 35321462
Grigorian N, Baumrucker SJ. Aromatase inhibitor-associated musculoskeletal pain: an overview of pathophysiology and treatment modalities. SAGE Open Med. 2022;10:20503121221078720.35321462
10. Hershman DL Shao T Kushi LH Early discontinuation and non-adherence to adjuvant hormonal therapy are associated with increased mortality in women with breast cancer Breast Cancer Res Treat 2011 126 2 529 537 10.1007/s10549-010-1132-4 20803066
Hershman DL, Shao T, Kushi LH, et al. Early discontinuation and non-adherence to adjuvant hormonal therapy are associated with increased mortality in women with breast cancer. Breast Cancer Res Treat. 2011;126(2):529–37.20803066
11. Hershman DL Lacchetti C Dworkin RH Prevention and management of chemotherapy-induced peripheral neuropathy in survivors of adult cancers: American Society of Clinical Oncology clinical practice guideline J Clin Oncol 2014 32 18 1941 1967 10.1200/JCO.2013.54.0914 24733808
Hershman DL, Lacchetti C, Dworkin RH, et al. Prevention and management of chemotherapy-induced peripheral neuropathy in survivors of adult cancers: American Society of Clinical Oncology clinical practice guideline. J Clin Oncol. 2014;32(18):1941–67.24733808
12. Mao JJ Ismaila N Bao T Integrative medicine for pain management in oncology: society for integrative oncology-ASCO guideline J Clin Oncol 2022 40 34 3998 4024 10.1200/JCO.22.01357 36122322
Mao JJ, Ismaila N, Bao T, et al. Integrative medicine for pain management in oncology: society for integrative oncology-ASCO guideline. J Clin Oncol. 2022;40(34):3998–4024.36122322
13. Swarm RA Paice JA Anghelescu DL Adult cancer pain, version 3.2019, NCCN clinical practice guidelines in oncology J Natl Compr Canc Netw. 2019 17 8 977 1007 10.6004/jnccn.2019.0038 31390582
Swarm RA, Paice JA, Anghelescu DL, et al. Adult cancer pain, version 3.2019, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw. 2019;17(8):977–1007.31390582
14. Colosia A Njue A Bajwa Z The burden of metastatic cancer-induced bone pain: a narrative review J Pain Res 2022 15 3399 3412 10.2147/JPR.S371337 36317162
Colosia A, Njue A, Bajwa Z, et al. The burden of metastatic cancer-induced bone pain: a narrative review. J Pain Res. 2022;15:3399–412.36317162
15. Broder MS Gutierrez B Cherepanov D Linhares Y Burden of skeletal-related events in prostate cancer: unmet need in pain improvement Support Care Cancer 2015 23 1 237 247 10.1007/s00520-014-2437-3 25270847
Broder MS, Gutierrez B, Cherepanov D, Linhares Y. Burden of skeletal-related events in prostate cancer: unmet need in pain improvement. Support Care Cancer. 2015;23(1):237–47.25270847
16. Hart NH Crawford-Williams F Crichton M Unmet supportive care needs of people with advanced cancer and their caregivers: a systematic scoping review Crit Rev Oncol Hemat. 2022 176 176 10.1016/j.critrevonc.2022.103728
Hart NH, Crawford-Williams F, Crichton M, et al. Unmet supportive care needs of people with advanced cancer and their caregivers: a systematic scoping review. Crit Rev Oncol Hemat. 2022;176:176.
17. Hershman DL Unger JM Greenlee H Comparison of acupuncture vs sham acupuncture or waiting list control in the treatment of aromatase inhibitor-related joint pain: a randomized clinical trial JAMA Netw Open 2022 5 11 e2241720 10.1001/jamanetworkopen.2022.41720 36367721
Hershman DL, Unger JM, Greenlee H, et al. Comparison of acupuncture vs sham acupuncture or waiting list control in the treatment of aromatase inhibitor-related joint pain: a randomized clinical trial. JAMA Netw Open. 2022;5(11).36367721
18. Mao JJ Liou KT Baser RE Effectiveness of electroacupuncture or auricular acupuncture vs usual care for chronic musculoskeletal pain among cancer survivors the PEACE randomized clinical trial Jama Oncol 2021 7 5 720 727 10.1001/jamaoncol.2021.0310 33734288
Mao JJ, Liou KT, Baser RE, et al. Effectiveness of electroacupuncture or auricular acupuncture vs usual care for chronic musculoskeletal pain among cancer survivors the PEACE randomized clinical trial. Jama Oncol. 2021;7(5):720–7.33734288
19. Musial F Acupuncture for the treatment of pain - a mega-placebo? Front Neurosci-Switz 2019 13 1110 10.3389/fnins.2019.01110
Musial F. Acupuncture for the treatment of pain - a mega-placebo? Front Neurosci-Switz. 2019;13:1110.
20. Fei YT Cao HJ Xia RY Methodological challenges in design and conduct of randomised controlled trials in acupuncture BMJ 2022 376 e064345 10.1136/bmj-2021-064345 35217507
Fei YT, Cao HJ, Xia RY, et al. Methodological challenges in design and conduct of randomised controlled trials in acupuncture. BMJ. 2022;376: e064345.35217507
21. Liou KT Ashare R Worster B SIO-ASCO guideline on integrative medicine for cancer pain management: implications for racial and ethnic pain disparities JNCI Cancer Spectr 2023 7 4 pkad042 10.1093/jncics/pkad042 37307074
Liou KT, Ashare R, Worster B, et al. SIO-ASCO guideline on integrative medicine for cancer pain management: implications for racial and ethnic pain disparities. JNCI Cancer Spectr. 2023;7(4):pkad042.37307074
22. He Y Guo X May BH Clinical evidence for association of acupuncture and acupressure with improved cancer pain: a systematic review and meta-analysis Jama Oncol 2020 6 2 271 278 10.1001/jamaoncol.2019.5233 31855257
He Y, Guo X, May BH, et al. Clinical evidence for association of acupuncture and acupressure with improved cancer pain: a systematic review and meta-analysis. Jama Oncol. 2020;6(2):271–8.31855257
23. Zeldow B Hatfield LA Confounding and regression adjustment in difference-in-differences studies Health Serv Res 2021 56 5 932 941 10.1111/1475-6773.13666 33978956
Zeldow B, Hatfield LA. Confounding and regression adjustment in difference-in-differences studies. Health Serv Res. 2021;56(5):932–41.33978956
24. Consumer price index (CPI) databases: U.S. Bureau of Labor Statistics. https://www.bls.gov/cpi/data.htm. Accessed 22 Feb 2024.
25. National Cancer Institute. SEER-Medicare: comorbidity SAS macros. 2021. https://healthcaredelivery.cancer.gov/seermedicare/considerations/calculation.html. Accessed 16 Jan 2023.
26. Quan H Sundararajan V Halfon P Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data Med Care 2005 43 11 1130 1139 10.1097/01.mlr.0000182534.19832.83 16224307
Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130–9.16224307
27. Charlson ME Pompei P Ales KL MacKenzie CR A new method of classifying prognostic comorbidity in longitudinal studies: development and validation J Chronic Dis 1987 40 5 373 383 10.1016/0021-9681(87)90171-8 3558716
Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373–83.3558716
28. DeClue RW Antol DD Casebeer AW Claims identification of patients with severe cancer-related symptoms Am J Manag Care 2020 26 6 e191 e197 10.37765/ajmc.2020.43495 32549069
DeClue RW, Antol DD, Casebeer AW, et al. Claims identification of patients with severe cancer-related symptoms. Am J Manag Care. 2020;26(6):e191–7.32549069
29. Ng DQ Chan D Acharya MM Grill JD Chan A Research attitude and interest among cancer survivors with or without cognitive impairment Cancers (Basel). 2023 15 13 3409 10.3390/cancers15133409 37444519
Ng DQ, Chan D, Acharya MM, Grill JD, Chan A. Research attitude and interest among cancer survivors with or without cognitive impairment. Cancers (Basel). 2023;15(13):3409.37444519
30. Beal SJ Kupzyk KA An introduction to propensity scores: what, when, and how J Early Adolescence 2014 34 1 66 92 10.1177/0272431613503215
Beal SJ, Kupzyk KA. An introduction to propensity scores: what, when, and how. J Early Adolescence. 2014;34(1):66–92.
31. Austin PC Stuart EA Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies Stat Med 2015 34 28 3661 3679 10.1002/sim.6607 26238958
Austin PC, Stuart EA. Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Stat Med. 2015;34(28):3661–79. 26238958
32. Cole SR Hernan MA Constructing inverse probability weights for marginal structural models Am J Epidemiol 2008 168 6 656 664 10.1093/aje/kwn164 18682488
Cole SR, Hernan MA. Constructing inverse probability weights for marginal structural models. Am J Epidemiol. 2008;168(6):656–64.18682488
33. Austin PC An introduction to propensity score methods for reducing the effects of confounding in observational studies Multivariate Behav Res 2011 46 3 399 424 10.1080/00273171.2011.568786 21818162
Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399–424.21818162
34. Zhang XW Hou WB Pu FL Acupuncture for cancer-related conditions: an overview of systematic reviews Phytomedicine 2022 106 154430 10.1016/j.phymed.2022.154430 36099656
Zhang XW, Hou WB, Pu FL, et al. Acupuncture for cancer-related conditions: an overview of systematic reviews. Phytomedicine. 2022;106: 154430. 36099656
35. Grant SJ, Kwon K, Naehrig D, Asher R, Lacey J. Characteristics and symptom burden of patients accessing acupuncture services at a cancer hospital. Integr Cancer Ther. 2021;20:15347354211002253.
36. Highfield ES Longacre M Chuang YH Burgess JF Does acupuncture treatment affect utilization of other hospital services at an urban safety-net hospital? J Altern Complem Med 2016 22 4 323 327 10.1089/acm.2015.0151
Highfield ES, Longacre M, Chuang YH, Burgess JF. Does acupuncture treatment affect utilization of other hospital services at an urban safety-net hospital? J Altern Complem Med. 2016;22(4):323–7.
37. Ho DE, Imai K, King G, Stuart EA. MatchIt: nonparametric preprocessing for parametric causal inference. J Stat Softw. 2011;42(8):1–28.
38. R Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2023. https://www.R-project.org/.
39. Nahin RL Rhee A Stussman B Use of complementary health approaches overall and for pain management by US adults Jama-J Am Med Assoc 2024 331 7 613 615 10.1001/jama.2023.26775
Nahin RL, Rhee A, Stussman B. Use of complementary health approaches overall and for pain management by US adults. Jama-J Am Med Assoc. 2024;331(7):613–5.
40. Birch S Bovey M Robinson N Acupuncture for chronic primary pain – are UK guidelines now consistent with other countries? Eur J Integr Med 2021 41 101257 10.1016/j.eujim.2020.101257
Birch S, Bovey M, Robinson N. Acupuncture for chronic primary pain – are UK guidelines now consistent with other countries? Eur J Integr Med. 2021;41: 101257.
41. Qaseem A Wilt TJ McLean RM Noninvasive treatments for acute, subacute, and chronic low back pain: a clinical practice guideline from the American College of Physicians Ann Intern Med 2017 166 7 514 530 10.7326/M16-2367 28192789
Qaseem A, Wilt TJ, McLean RM, et al. Noninvasive treatments for acute, subacute, and chronic low back pain: a clinical practice guideline from the American College of Physicians. Ann Intern Med. 2017;166(7):514–30.28192789
42. Peets JM Pomeranz B Cxbk mice deficient in opiate receptors show poor electroacupuncture analgesia Nature 1978 273 5664 675 676 10.1038/273675a0 208000
Peets JM, Pomeranz B. Cxbk mice deficient in opiate receptors show poor electroacupuncture analgesia. Nature. 1978;273(5664):675–6.208000
43. Pomeranz B Chiu D Naloxone blockade of acupuncture analgesia: endorphin implicated Life Sci 1976 19 11 1757 1762 10.1016/0024-3205(76)90084-9 187888
Pomeranz B, Chiu D. Naloxone blockade of acupuncture analgesia: endorphin implicated. Life Sci. 1976;19(11):1757–62.187888
44. Goldman N Chen M Fujita T Adenosine A1 receptors mediate local anti-nociceptive effects of acupuncture Nat Neurosci 2010 13 7 883 888 10.1038/nn.2562 20512135
Goldman N, Chen M, Fujita T, et al. Adenosine A1 receptors mediate local anti-nociceptive effects of acupuncture. Nat Neurosci. 2010;13(7):883–8.20512135
45. Baumler P Zhang W Stubinger T Irnich D Acupuncture-related adverse events: systematic review and meta-analyses of prospective clinical studies BMJ Open 2021 11 9 e045961 10.1136/bmjopen-2020-045961 34489268
Baumler P, Zhang W, Stubinger T, Irnich D. Acupuncture-related adverse events: systematic review and meta-analyses of prospective clinical studies. BMJ Open. 2021;11(9): e045961.34489268
46. Mao JJ Oncology acupuncture: precision medicine meets patient-centered care Med Acupunct 2020 32 6 405 407 10.1089/acu.2020.1501 33362901
Mao JJ. Oncology acupuncture: precision medicine meets patient-centered care. Med Acupunct. 2020;32(6):405–7.33362901
47. Lagendijk M Mittendorf E King TA Gibbons C Pusic A Dominici LS Incorporating patient-reported outcome measures into breast surgical oncology: advancing toward value-based care Oncologist 2020 25 5 384 390 10.1634/theoncologist.2019-0355 31848315
Lagendijk M, Mittendorf E, King TA, Gibbons C, Pusic A, Dominici LS. Incorporating patient-reported outcome measures into breast surgical oncology: advancing toward value-based care. Oncologist. 2020;25(5):384–90.31848315
48. Judson PL Abdallah R Xiong Y Ebbert J Lancaster JM Complementary and alternative medicine use in individuals presenting for care at a comprehensive cancer center Integr Cancer Ther 2017 16 1 96 103 10.1177/1534735416660384 27444312
Judson PL, Abdallah R, Xiong Y, Ebbert J, Lancaster JM. Complementary and alternative medicine use in individuals presenting for care at a comprehensive cancer center. Integr Cancer Ther. 2017;16(1):96–103. 27444312
49. Austin S Ramamonjiarivelo Z Qu H Ellis-Griffith G Acupuncture use in the United States: who, where, why, and at what price? Health Mark Q 2015 32 2 113 128 10.1080/07359683.2015.1033929 26075541
Austin S, Ramamonjiarivelo Z, Qu H, Ellis-Griffith G. Acupuncture use in the United States: who, where, why, and at what price? Health Mark Q. 2015;32(2):113–28.26075541
50. Baker K McDonald J Steel A Tackling health inequity: a commentary on the potential of acupuncture to improve health outcomes of marginalised populations Acupunct Med 2021 39 5 533 537 10.1177/0964528420961404 33280395
Baker K, McDonald J, Steel A. Tackling health inequity: a commentary on the potential of acupuncture to improve health outcomes of marginalised populations. Acupunct Med. 2021;39(5):533–7.33280395
51. Huemer M Pansi A Hofmann G Symptom clusters associated with complementary and alternative medicine use by cancer patients: a cross-sectional study Plos One 2023 18 12 e0294641 10.1371/journal.pone.0294641 38096236
Huemer M, Pansi A, Hofmann G, et al. Symptom clusters associated with complementary and alternative medicine use by cancer patients: a cross-sectional study. Plos One. 2023;18(12):e0294641.38096236
52. Wolever RQ Goel NS Roberts RS Integrative medicine patients have high stress, pain, and psychological symptoms Explore-Ny 2015 11 4 296 303 10.1016/j.explore.2015.04.003 26044918
Wolever RQ, Goel NS, Roberts RS, et al. Integrative medicine patients have high stress, pain, and psychological symptoms. Explore-Ny. 2015;11(4):296–303.26044918
53. Lam CS Ma CT Li MCH The use of traditional, complementary and integrative medicine in Chinese adolescent and young adult patients with cancer: a multicenter cross-sectional study Eur J Oncol Nurs 2023 66 102377 10.1016/j.ejon.2023.102377 37499405
Lam CS, Ma CT, Li MCH, et al. The use of traditional, complementary and integrative medicine in Chinese adolescent and young adult patients with cancer: a multicenter cross-sectional study. Eur J Oncol Nurs. 2023;66: 102377.37499405
54. Centers for Medicare & Medicaid Services. Acupuncture for chronic low back pain (CAG-00452N) - decision memo. https://www.cms.gov/medicare-coverage-database/view/ncacal-decision-memo.aspx?proposed=N&NCAId=295. Published 2020. Updated January 21, 2020. Accessed 7 March 2024.
55. Enewold L Parsons H Zhao L Updated overview of the SEER-Medicare data: enhanced content and applications J Natl Cancer Inst Monogr 2020 2020 55 3 13 32412076
Enewold L, Parsons H, Zhao L, et al. Updated overview of the SEER-Medicare data: enhanced content and applications. J Natl Cancer Inst Monogr. 2020;2020(55):3–13.32412076
56. Mues KE Liede A Liu J Use of the Medicare database in epidemiologic and health services research: a valuable source of real-world evidence on the older and disabled populations in the US Clin Epidemiol 2017 9 267 277 10.2147/CLEP.S105613 28533698
Mues KE, Liede A, Liu J, et al. Use of the Medicare database in epidemiologic and health services research: a valuable source of real-world evidence on the older and disabled populations in the US. Clin Epidemiol. 2017;9:267–77.28533698
