Prolonged Opioid Use After Surgery for Early‐Stage Breast Cancer

Abstract

Introduction

This study examined the patterns of prolonged opioid use and the factors associated with higher risk of prolonged opioid use among opioid‐naïve working‐age patients with early‐stage breast cancer.

Methods

Using MarketScan data, the study identified 23,440 opioid‐naïve patients who received surgery for breast cancer between January 2000 and December 2014 and filled at least one opioid prescription attributable to surgery. Prolonged opioid use was defined as one or more prescriptions for opioids within 90 to 180 days after surgery and defined extra‐prolonged opioid use as one or more opioid prescriptions between 181 and 365 days after surgery. Multivariable logistic regressions were performed to ascertain factors associated with prolonged and extra‐prolonged use of opioids.

Findings

Of the 23,440 patients, 4,233 (18%) had prolonged opioid use, and 2,052 (9%) had extra‐prolonged opioid use. Patients who received mastectomy plus reconstruction had the highest rate of prolonged opioid use (38%) followed by mastectomy alone (15%). A multivariable logistic regression confirmed that patients with mastectomy and reconstruction had the highest odds ratio of prolonged opioid use compared to lumpectomy and whole breast irradiation (adjusted odds ratio, 5.6; 95% confidence interval, 5.1–6.1). Mean daily opioid dose was consistently high without any obvious dosage reduction among patients with opioid use.

Interpretation

This large observational study showed a high rate of prolonged opioid use among patients who received surgery for early‐stage breast cancer and found significant difference in prolonged opioid use by treatment type.

Implications for Practice

This large observational study found a high rate of prolonged opioid use among working‐age patients with early‐stage breast cancer who received curative surgery, especially among patients who received mastectomy. Among patients with opioid use, the mean daily opioid dose was consistently high without any obvious dosage tapering. This study highlights the need to emphasize appropriate opioid therapy and potential dosage reduction or discontinuation among patients with early‐stage breast cancer who received surgical interventions.

Short abstract

Although pain management is a critical part of patient care, unnecessary prolonged opioid use is associated with adverse consequences. This article examines prolonged opioid use patterns among patients with cancer, comparing different treatment modalities and identifying factors associated with prolonged opioid use.

Introduction

Opioid medications are often prescribed for postsurgical analgesia, but their routine use has come under scrutiny as the U.S. continues to be embroiled in the opioid crisis 1. In 2017, more than 11 million Americans misused prescription opioids in the past year 2. Although pain management is a critical part of patient care 3, unnecessary prolonged opioid use is associated with adverse physiological, psychological, and economic consequences 4, 5, 6, 7, 8. In 2013, the annual cost of opioid misuse was estimated at $78.5 billion 9.

Although patients with cancer generally require surgical treatment to survive early‐stage cancer, they are also more susceptible to prolonged opioid use—thereby putting them at greater risk for adverse consequences 10. Clinicians have guidelines for opioid prescribing to treat pain associated with advanced cancer disease 11, but they are still lacking in guidance to treat early‐stage cancer. Opioid prescribing practices after surgical intervention commonly leave patients with unused opioid medication long after the time needed for analgesia; however, current research has primarily focused on patients without cancer 12, 13, 14, 15. For example, nearly three quarters of patients undergoing orthopedic surgery had leftover opioid medications after discontinuing therapy 15. Patients with unused opioid medications are at a greater risk of misusing them 16. With this increased risk, the focus must now also include patients being treated for early‐stage breast cancer with surgical interventions.

Breast cancer is the most common cancer in women, and surgical resection of the tumor is a standard component of curative therapy for localized, early‐stage disease 17, 18, 19. Options for local therapy for early‐stage breast cancer include mastectomies (Mast alone), mastectomies followed by reconstruction (Mast+Recon), lumpectomies with whole breast irradiation (Lump+WBI), and, in a subgroup of early cancers with favorable biology, lumpectomies with brachytherapy (Lump+Brachy) or lumpectomies without radiation (Lump alone) 20. These treatments differ in terms of both cost and complications, including the potential for pain and the degree of psychological distress for the patient 18. Although the 5‐year survival rate for newly diagnosed breast cancer approaches 90% 19, prolonged opioid use after treatment should be assessed among this patient population to prevent prolonged opioid exposure as well as the associated risk for long‐term life‐altering adverse consequences.

There have been many studies comparing the survival outcomes and economic costs of different surgical interventions for breast cancer; however, there has been no research on the associated prolonged opioid use. Therefore, this large observational study was conducted to assess the prolonged opioid use among patients with localized, early‐stage breast cancer who underwent different modalities of curative surgery. This study aims to examine prolonged opioid use patterns among these patients by comparing different treatment modalities and identifying factors associated with prolonged opioid use.

Materials and Methods

Data Source

The study used the IBM Truven Health MarketScan database. This claims‐based longitudinal database covers 50 million unique patients who are enrolled in commercial health insurance plans sponsored by over 100 large or medium‐sized U.S.‐based employers. It includes various different types of health insurance plans, including health maintenance organizations (HMOs), preferred provider organizations, point‐of‐service (POS) plans, and indemnity plans, and a variety of coverage, such as privately insured fee‐for‐service, POS, or capitated health plans. The geographical coverage is representative of the U.S. population, and this database is a well‐accepted data source for treatment and outcomes research 21, 22, 23, 24.

Study Cohort

The study identified female patients aged 18 to 64 years, diagnosed with incident, early‐stage breast cancer from January 2000 to December 2014 based on a validated claims‐based algorithm 25. It was required that the patients have continuous insurance coverage from 12 months prior to and 12 months after the diagnosis date in order to ensure complete records for the identification of prediagnosis comorbidities, postdiagnosis treatments, and incident opioid use. This study focused on opioid‐naïve patients who received one of the five local treatment options we studied within 12 months of diagnosis (i.e., Mast only, Mast+Recon, Lump only, Lump+WBI, and Lump+Brachy) and filled at least one opioid prescription between 30 days before and 14 days after discharge. It was also required that the patients have 12 months’ continuous insurance coverage after their surgery procedure to ensure that we captured the potential opioid prescribing pattern during this time frame. Patients were considered to be opioid‐naïve if they did not have any opioid prescriptions between the 365 and 30 days before surgery. The study further excluded patients who had multiple surgeries, had a length of stay above 30 days, discharged to hospice, or died in the hospital, as these patients probably had more severe disease. In this paper, we aim to focus on only patients with localized, early‐stage breast cancer undergoing single‐stage surgery as the risk of prolonged opioid use would presumably increase with each subsequent surgery.

The detailed inclusion and exclusion criteria are provided in supplemental online Figure 1.

Outcome Variables

The key outcome variable was prolonged opioid use. Prolonged opioid use was defined as one or more opioid prescriptions within 1 to 90 days after surgery discharge along with one or more prescriptions for opioids within 91 to 180 days after discharge, which is a widely accepted criteria in the literature 10, 26, 27, 28. Furthermore, we also measured the use of opioid prescriptions between 181 and 365 days after surgery to explore the even longer term opioid use, which was defined as extra‐prolonged opioid use in our study. Opioid prescription dose was converted to oral morphine equivalents (OMEs) and obtained from pharmacy claims 29. Another key outcome variable was the mean daily opioid dose (OME) during the 12 months after surgery, which was calculated every month by multiplying the OME conversion factor by prescribed quantity then dividing by the days supplied.

Independent Variables

The main independent variable was type of surgical/local therapy intervention. We had five treatment groups in our study—Mast only, Mast+Recon, Lump only, Lump+WBI, and Lump+Brachy—using the same algorithm as in the study by Smith et al. 18. The Lump+WBI group was limited to those patients who received at least 15 unique external beam radiation treatments without concomitant brachytherapy. The Lump+Brachy group was limited to patients who received brachytherapy without concomitant external beam radiation. The Mast+Recon group was limited to patients with mastectomy within 1 year of diagnosis and a code for breast reconstruction on the same day of mastectomy.

In addition to the main independent variable, other covariates included patients’ sociodemographic and clinical factors. The sociodemographic covariates included year of diagnosis (2000–2004, 2005–2009, and 2010–2014), age in years at diagnosis (<40, 40–49, 50–59, 60–64), type of coverage (non‐HMO, HMO), and covered individual (employee, dependent). The clinical covariates included Charlson comorbidity scores (zero, one, or at least two), binary variables (yes/no) indicating psychiatric conditions (depression, anxiety, severe mental illness, and substance use disorder), a binary variable (yes/no) indicating receiving chemotherapy, a binary variable (yes/no) indicating having positive node, and total opioid use (OME) during perioperative period (categorized into quartiles: ≤150, 151–225, 226–450, >450). For the Charlson comorbidity score, this study used the Deyo‐Romano modified Charlson comorbidity score, which is a commonly adopted measure for ascertainment of comorbidity in studies using claims data 30, 31, 32, 33. The comorbidity score and psychiatric conditions were derived from insurance claims during the 12 months preceding diagnosis.

Analyses

Group differences in prolonged opioid use and extra‐prolonged opioid use were tested with chi‐squared statistics. The study used two multivariable logistic regressions to estimate whether different treatment type was associated with either prolonged opioid use or extra‐prolonged opioid use and presented findings as adjusted odds ratios (AORs) and their corresponding 95% confidence intervals (CIs). All statistical analyses were conducted in SAS 9.3 (SAS Institute, Cary, NC). The Institutional Review Board at The University of Texas MD Anderson Cancer Center exempted this study from review because all patients in the database had been deidentified.

Results

Characteristics of the study cohort are provided in Table 1. Of 23,440 patients identified with breast cancer, 4,233 (18.06%) had prolonged opioid use, and 2,052 (8.75%) had extra‐prolonged opioid use. Among the treatment groups, patients with mastectomy plus breast reconstruction had the highest rates of prolonged opioid use (38.32%) compared with lumpectomy plus whole breast irradiation (7.88%), lumpectomy plus brachytherapy (6.59%), lumpectomy alone (4.75%), and mastectomy alone (14.53%) groups. Furthermore, the prolonged opioid use rate was higher among younger patients, patients having non‐HMO coverage, severe mental illness, depression, anxiety, substance use disorder, chemotherapy, node positive, and greater opioid use during perioperative period. In terms of the extra‐prolonged opioid use, the result was similar except that the type of coverage was no longer significant. For instance, the group with mastectomy plus breast reconstruction still had the highest rates of prolonged opioid use (18.94%) compared with lumpectomy plus whole breast irradiation (3.75%), lumpectomy plus brachytherapy (3.36%), lumpectomy alone (2.59%), and mastectomy alone (6.3%) groups.

Table 1.

Patient characteristics by prolonged and extra‐prolonged opioid use in the study cohort of opioid‐naïve patients (n = 23,440)

Characteristics	Naïve patients with prolonged opioid use (n = 4,233), n (%)	p value	Naïve patients with extra‐prolonged opioid use (n = 2,052), n (%)	p value
Type of surgery		<.0001		<.0001
Lump+WBI	800 (7.88)		381 (3.75)
Lump+Brachy	94 (6.59)		48 (3.36)
Lump alone	55 (4.75)		30 (2.59)
Mast alone	498 (14.53)		216 (6.3)
Mast+Recon	2,786 (38.32)		1,377 (18.94)
Year of diagnosis		.0097		.0082
2000–2004	416 (15.98)		192 (7.37)
2005–2009	1,512 (18.6)		758 (9.32)
2010–2014	2,305 (18.14)		1,102 (8.67)
Age		<.0001		<.0001
<40	432 (31.3)		206 (14.93)
40–49	1,414 (21.53)		675 (10.28)
50–59	1,753 (16.44)		842 (7.9)
60–64	634 (13.13)		329 (6.81)
Type of insurance		.0002		.1314
Non‐HMO	3,579 (18.48)		1,720 (8.88)
HMO	654 (16.05)		332 (8.15)
Covered individual		.24		.42
Employee	1,736 (18.42)		842 (8.93)
Dependent	2,497 (17.82)		1,210 (8.63)
Comorbidity		.78		.0656
0	3,869 (18.02)		1,856 (8.64)
1	317 (18.34)		167 (9.66)
>2	47 (19.58)		29 (12.08)
Severe mental illness		.03		.0015
Yes	51 (23.61)		32 (14.81)
No	4,182 (18.01)		2,020 (8.7)
Depression		<.0001		<.0001
Yes	357 (24.27)		203 (13.8)
No	3,876 (17.64)		1,849 (90.11)
Anxiety		<.0001		<.0001
Yes	267 (23.04)		151 (13.03)
No	3,966 (17.8)		1,901 (8.53)
Substance use disorder		.001		.0016
Yes	96 (24.37)		52 (13.2)
No	4,137 (17.95)		2,000 (8.68)
Chemotherapy		<.0001		<.0001
Yes	1,916 (22.41)		1,060 (12.4)
No	2317 (15.56)		992 (6.66)
Node positive		<.0001		<.0001
Yes	697 (24.52)		368 (12.95)
No	3,536 (17.17)		1,684 (8.18)
Total opioid use (OMEs) during perioperative period		<.0001		<.0001
≤150 mg	665 (8.45)		278 (3.53)
151–225 mg	588 (12.41)		254 (5.36)
226–450 mg	1,376 (32.51)		660 (11.27)
>450 mg	1,604 (37.89)		860 (17.30)

Abbreviations: Brachy, brachytherapy; HMO, health maintenance organization; Lump, lumpectomy; Mast, mastectomy; OME, oral morphine equivalent; Recon, reconstruction; WBI, whole breast irradiation.

In both multivariable logistic regression models, surgical intervention type was still strongly associated with both prolonged and extra‐prolonged use. For prolonged use, compared with the lumpectomy plus whole breast irradiation group, the lumpectomy alone group was less likely to have prolonged use with an AOR of 0.65 (95% CI, 0.49–0.86), whereas mastectomy alone and mastectomy plus breast reconstruction groups were more likely to have prolonged use with an AOR of 1.66 (95% CI, 1.47–1.88) and AOR of 5.56 (95% CI, 5.06–6.1), respectively. The results also showed significant associations between year of diagnosis, age, type of coverage, comorbidity scores, depression, anxiety, substance use disorder, chemotherapy, node positive, total opioid use during perioperative period, and prolonged use. For extra‐prolonged use, compared with lumpectomy plus whole breast irradiation group, the mastectomy alone and mastectomy plus breast reconstruction groups were more likely to have extra‐prolonged use with an AOR of 1.35 (95% CI, 1.13–1.61) and AOR of 4.51 (95% CI, 3.97–5.13), respectively. We found that prior psychiatric conditions, including depression, anxiety, and substance use disorder, were significantly associated with higher likelihood of prolonged and extra‐prolonged opioid use. Other significant associations were found between diagnosis, comorbidity scores, chemotherapy, node positive, total opioid use during perioperative period, and prolonged use. Detailed results are provided in Table 2. We also explored the interactions between prior psychiatric conditions and surgical interventions and found no significant interactions. It seems that the impact of psychiatric conditions on opioid use does not significantly vary by surgical modes.

Table 2.

Logistic regression results for prolonged and extra‐prolonged opioid use

	Prolonged opioid use		Extra‐prolonged opioid use
	Odds ratio (95% CI)	p value	Odds ratio (95% CI)	p value
Type of surgery (ref = Lump+WBI)
Lump+Brachy	0.896 (0.716–1.122)	.3386	0.995 (0.73–1.356)	.9763
Lump alone	0.65 (0.489–0.863)	.0029	0.842 (0.575–1.233)	.3773
Mast alone	1.664 (1.47–1.884)	<.0001	1.345 (1.126–1.606)	.0011
Mast+Recon	5.555 (5.061–6.097)	<.0001	4.513 (3.973–5.127)	<.0001
Year of diagnosis (ref = 2000–2004)
2005–2009	1.244 (1.093–1.416)	.0009	1.323 (1.112–1.574)	.0016
2010–2014	1.159 (1.021–1.315)	.0223	1.199 (1.011–1.421)	.0369
Age (ref = '<40')
40–49	0.862 (0.75–0.992)	.0377	0.918 (0.769–1.096)	.3435
50–59	0.847 (0.738–0.972)	.0181	0.957 (0.803–1.14)	.6202
60–64	0.757 (0.647–0.886)	.0005	0.978 (0.8–1.195)	.8292
Type of insurance (ref = non‐HMO)
HMO	0.807 (0.731–0.892)	<.0001	0.897 (0.788–1.021)	.0999
Covered individual (ref = Dependent)
Employee	1.022 (0.948–1.1)	.5735	1.004 (0.911–1.107)	.9315
Comorbidity (ref = 0)
1	1.242 (1.08–1.428)	.0024	1.324 (1.108–1.583)	.002
≥2	1.642 (1.16–2.324)	.0052	2.069 (1.364–3.139)	.0006
Severe mental illness (ref = No)
Yes	1.143 (0.804–1.624)	.4565	1.425 (0.946–2.148)	.09
Depression (ref = No)
Yes	1.346 (1.167–1.552)	<.0001	1.499 (1.261–1.781)	<.0001
Anxiety (ref = No)
Yes	1.206 (1.027–1.416)	.0224	1.376 (1.131–1.675)	.0014
Substance use disorder (ref = No)
Yes	1.489 (1.148–1.932)	.0027	1.493 (1.084–2.055)	.0141
Chemotherapy (ref = No)
Yes	1.46 (1.348–1.581)	<.0001	1.931 (1.742–2.14)	<.0001
Node positive (ref = No)
Yes	1.285 (1.153–1.433)	<.0001	1.193 (1.041–1.367)	.0109
Total opioid use (OMEs) during perioperative period (ref = '≤150 mg')
151–225 mg	1.202 (1.063–1.36)	.0033	1.243 (1.04–1.485)	.0167
226–450 mg	1.848 (1.659–2.058)	<.0001	2.026 (1.74–2.36)	<.0001
>450 mg	2.606 (2.338–2.904)	<.0001	3.042 (2.618–3.536)	<.0001

Abbreviations: Brachy, brachytherapy; CI, confidence interval; HMO, health maintenance organization; Lump, lumpectomy; Mast, mastectomy; OME, oral morphine equivalent; Recon, reconstruction; WBI, whole breast irradiation.

Figure 1A shows the proportion of patients taking opioid drug during the 12 months after surgery stratified by treatment type. All the treatment groups had a sharp drop in the second month after surgery, after which the proportion was gradually decreasing except for the group with mastectomy plus reconstruction. Furthermore, this group had much higher proportion of opioid use than other treatment groups throughout the 12 months.

Figure 1.

Opioid use during the year after surgery. (A): Proportion of patients who filled opioid prescriptions during the year after surgery. (B): Mean daily opioid use among patients who filled opioid prescriptions during the year after surgery.

Abbreviations: Brachy, brachytherapy; Lump, lumpectomy; Mast, mastectomy; OME, oral morphine equivalent; Recon, reconstruction; WBI, whole breast irradiation.

Figure 1B shows the trend of mean daily opioid dose 12 months after surgery among patients who took opioids. Among these opioid users, the mean daily opioid for all the treatment groups was consistently high (ranging from 35 mg OME to 96 mg OME) during the 12 months after surgery. We do not observe any obvious dosage tapering within any group as the mean daily opioid dose did not significantly decrease as the time from surgical intervention date grew.

Discussion

To the best of our knowledge, this is the first large observational study focusing on prolonged opioid use among patients with early‐stage breast cancer who underwent different surgical treatments. We found that nearly one fifth of our early‐stage breast cancer cohort was prescribed opioids for up to 180 days and that one in 11 were prescribed opioids for up to 1 year. With newly diagnosed cases of breast cancer having increased to over a quarter million cases annually, the prevention of prescription opioid misuse is a growing issue among breast cancer survivors.

The study showed that patients with breast cancer receiving treatment with mastectomy plus reconstruction were significantly more likely to be prescribed an opioid medication for a prolonged duration. This surgical intervention type accounted for approximately one third of our early‐stage breast cancer cohort. If this proportion is applied to the newly diagnosed breast cancer cases in 2018 19, then approximately 83,000 patients annually would be at a significantly higher risk of prolonged opioid use and its associated adverse consequences. Our data demonstrate that patients with mastectomy, both with and without reconstruction, are much more likely to have prolonged opioid use compared with patients treated with breast conserving treatment (lumpectomy). As mastectomy is a more extensive surgery, a greater need for pain management is expected, physicians are more likely to prescribe opioids for management of pain, and the patients are more likely to use the opioids prescribed. Because patients treated with a mastectomy, irrespective of reconstruction after, are at the highest risk of both prolonged and extra‐prolonged opioid prescribing, they should be provided additional monitoring strategies aimed at preventing prescription opioid misuse. This study focused on patients with early‐stage breast cancer who are at working age and are likely be long‐term survivors. These patients are thus most vulnerable to potential long‐term adverse effects of prolonged opioid use. Appropriate pain management is critical for these working‐age patients who are very likely to be long‐term survivors. Furthermore, the current trend of growing use of contralateral prophylactic mastectomy, especially prevalent among privately insured and younger patients 34, 35 makes this study especially relevant and highlights the urgent need for strategies to manage pain appropriately and prevent unnecessary long‐term use of opioids in this group of patients. Our study also showed that prior psychiatric conditions, including depression, anxiety, and substance use disorder, are associated with a significantly higher likelihood of prolonged and extra‐prolonged opioid use. This is in line with many other studies in the literature showing a significant connection between prior psychiatric conditions and postoperative pain and opioid use 36, 37, 38, 39, 40. Therefore, it is imperative that physicians consider their patients’ psychiatric history when prescribing pain medications. A prior history of either alcohol or tobacco use is also a potential risk factor for prolonged opioid use 26, 41; however, such information was not available in the database that we used for this study.

One method that has been shown to reduce the incidence of prolonged opioid involves steadily reducing dosage until complete discontinuation of opioid therapy. This method, often called tapering or dose de‐escalation, is recommended when the patient is not having a clinically meaningful improvement in pain and function 42. In the 12 months after surgical intervention, our study cohort did not show any obvious dose tapering among patients who were prescribed opioids for a prolonged duration. Tapering can be difficult and time consuming and thus may not be a priority for either the clinician or the patient who is recovering from a cancer surgery. However, to reduce the long‐term potential for harm associated with prolonged opioid prescribing, a risk mitigation plan including monitoring and tapering needs to be considered by the patient and treating clinicians. Another strategy is to replace the use of prescription opioid with alternative therapies such as regional anesthesia, infiltration of long‐acting anesthetics, regional nerve blocks, acetaminophen, nonsteroidal anti‐inflammatory drugs, gabapentin, and muscle relaxant drugs especially for patients with submuscular implants. More research is needed to evaluate the efficacy of various alternative strategies 43. Recent development of transitional pain service models involving a team of physicians, nurse practitioners, pharmacists, physical therapists, and psychologists could be a valuable asset to reduce the risk of opioid overuse/misuse targeting high‐risk patients who undergo mastectomy 44, 45.

As the U.S. is in the midst of the opioid crisis, broad policy changes are making both the prescribing and filling of opioid medications more difficult. The study findings were in line with the study by Lee et al. that identified incident long‐term opioid use as a common problem in patients undergoing curative‐intent surgery for a broad range of cancer types and recommended prescribing guidelines and increased patient counseling 10. The rate of prolonged opioid use seems to be considerably higher among patients with cancer who received surgical treatment when compared with patients who received noncancer surgeries overall. One recent systematic review and meta‐analysis on prolonged opioid use after trauma or surgery found the rate of prolonged opioid use to be approximately 2% among opioid‐naïve patients 36, which is significantly lower than the 18% in our study. However, the same paper also showed a much higher rate of prolonged opioid use of 16% among trauma patients 36, which is close to the 18% rate in this study. Such results are consistent with our finding that the rate of prolonged opioid use is highly associated with the type of surgery. The need for a patient‐centered focus is critical because a patient's recovery and transition to cancer survivorship is unique. Taking into account the finding that prolonged opioid use is associated with surgery type, a uniform reduction in opioid prescribing after surgery may not be prudent. Future research can focus on early‐stage breast cancer survivors who undergo surgery interventions, emphasizing the need for prescribers to be educated regarding best practices for opioid prescribing, especially appropriate tapering and potentially discontinuation of opioid therapy to prevent addiction 46. The study of persistent postmastectomy pain is beyond the scope of the current study. However, our findings suggest that many patients with breast cancer, especially those treated with mastectomy, may have prolonged pain. One systematic review about postsurgery pain in patients with breast cancer found that persistent pain after breast cancer treatment is prevalent at around 30% overall, and in a pooled sample of patients with pain severity information, 19% reported mild pain and 22% reported moderate to severe pain 47. Another meta‐analysis study on persistent pain after breast surgery found a median prevalence of 37.5%; furthermore, this study showed that factors such as radiotherapy, younger age, and axillary lymph node dissection were associated persistent pain 48. It has been shown that younger patients are more likely to develop persistent pain after breast cancer surgery, whereas older patients often have decreased pain perception and sensitivity 48, 49. It is possible that radiotherapy and chemotherapy induce neuropathy. Therefore, it is important for clinicians to take into account all these factors potentially associated with persistent pain.

Strengths of this study include following patients across multiple providers and settings using commercial claims data. The longitudinal design allowed for a monthly assessment of opioid prescribing during the 12 months after surgery. Also, this design allowed the study cohort to only include patients who were opioid‐naïve. This study also has some potential limitations that are common to claims‐based observational studies. For example, only opioid claims were assessed, as actual medication use cannot be captured in claims data. The administrative database lacks detailed clinical information such as the size of the tumor, degree of differentiation, biological markers, etc. The MarketScan data do not have information on pain, social capital, medication beliefs, and response to treatment, which may affect the likelihood of prolonged or extra‐prolonged opioid use. Other than the coded mental health diagnoses, we do not have detailed information on patients’ psychophysical and psychosocial profile, which may be highly associated with persistent pain after breast cancer surgery 48, 50. Some future directions include examining the consequences of chronic opioid therapy and the use of alternative pain management strategies other than opioids.

Conclusion

Prolonged opioid use can have significant adverse consequences for patients with early‐stage breast cancer being treated with surgical intervention. The likelihood of being prescribed opioids for long periods of time after surgery is affected by the type of surgical intervention performed. It is especially important to place an emphasis on reducing unnecessary prolonged prescribing of opioids in patients who undergo surgery for early‐stage breast cancer.

Author Contributions

Conception/design: Chan Shen, J. Douglas Thornton, Sharon H. Giordano

Provision of study material or patients: Sharon H. Giordano

Collection and/or assembly of data: Chan Shen, Dian Gu

Data analysis and interpretation: Chan Shen, J. Douglas Thornton, Dian Gu, Daleela Dodge, Shouhao Zhou, Weiguo He, Hui Zhao, Sharon H. Giordano

Manuscript writing: Chan Shen, J. Douglas Thornton, Dian Gu, Daleela Dodge, Shouhao Zhou, Weiguo He, Hui Zhao, Sharon H. Giordano

Final approval of manuscript: Chan Shen, J. Douglas Thornton, Dian Gu, Daleela Dodge, Shouhao Zhou, Weiguo He, Hui Zhao, Sharon H. Giordano

Disclosures

The authors indicated no financial relationships.

Supporting information

See http://www.TheOncologist.com for supplemental material available online.

Supplementary Figure. Derivation of the study cohort