Routine Opioid Prescriptions Are Not Necessary After Breast Excisional Biopsy or Lumpectomy Procedures

Abstract

Background

Opioid analgesics are overprescribed after surgery. In 8/2018 we replaced routine discharge opioid prescription with an NSAID for patients having lumpectomy or excisional biopsy (lump/ex). Here we compare patient-reported post-discharge pain scores among patients treated before/after the routine discharge medication change.

Methods

Patients were categorized based on treatment before/after discharge medication change: study period 1, routine opioids; study period 2, routine NSAID. Pain severity was assessed with an electronic survey on postoperative days (PODs) 1–5. Multivariable generalized estimating equations tested association between pain severity and discharge in the first versus second study period.

Results

1606 patients had lump/ex between 12/2017–6/2019: 789 (49%) reported pain scores and were analyzed (328 first period, 461 second period). Opioid prescription at discharge decreased from 96% in the first study period to 14% (95% CI 11%−18%) in the second. Only 1% of patients discharged with NSAID were later prescribed an opioid. Maximum reported pain score on any POD for all patients was: severe 3.8% (30); moderate 28% (217); mild 54% (430); none 14% (112). The estimated risk of moderate/greater pain on POD 1 was 36% and 34% for the first and second study periods, respectively. The proportion of patients reporting moderate/greater pain was non-significantly lower in patients treated in the second period (OR 0.91; 95% CI 0.67–1.22; p=0.5).

Conclusions

In patients undergoing lump/ex, we can exclude a clinically meaningful difference in reported post-discharge pain scores with a change to routine NSAID at discharge. Patients undergoing lump/ex should not be routinely discharged with opioids.

In 2017 there were 47,600 deaths from drug overdose in the United States; of these, 17,029 were attributed to prescription opioids.¹ A plethora of studies have demonstrated that opioids are overprescribed in the postoperative setting, and that unused tablets become available for diversion and misuse in the community.^2–9 There are now nationwide initiatives to reduce the use of prescription opioids for acute pain management.^10–13 The success of these initiatives hinges on a re-evaluation of current prescribing practices and investigation into pain management strategies that reduce unnecessary opioid prescription without compromising pain control. Lumpectomy and excisional biopsy are commonly performed ambulatory breast surgical procedures for which patients have been routinely discharged with opioid analgesics. Based on our observation that these procedures are generally not associated with a significant amount of pain after discharge, we changed our Enhanced Recovery After Surgery (ERAS) protocol in August 2018, from routine discharge with opioids, to discharge with a non-steroidal anti-inflammatory drug (NSAID). In this study, we retrospectively evaluated frequency of opioid prescription, and compared patient-reported post-discharge pain scores among patients treated before and after the change in routine discharge medication.

METHODS

Institutional review board approval was obtained with waiver of informed consent for retrospective analysis. We retrospectively identified patients undergoing lumpectomy (partial mastectomy; current procedural terminology [CPT] code 19301) and excisional biopsy procedures (CPT codes 19125 and 19120) at the Josie Robertson Surgery Center, a free-standing ambulatory surgery facility of Memorial Sloan Kettering Cancer Center, from December 2017 through June 2019. Patients having an axillary procedure (sentinel node biopsy or axillary dissection) were not included.

At our institution, all ambulatory breast surgery patients are routinely treated on an ERAS protocol, which includes intraoperative acetaminophen and ketorolac administration, and use of both long- and short-acting local anesthetic. The method of long-acting local anesthetic (bupivacaine) administration is based on surgeon preference, and is typically injected into the subcutaneous tissue and breast parenchyma surrounding the surgical cavity at the end of the case, or injected as a mixture with lidocaine into the skin, subcutaneous tissue, and breast parenchyma at the beginning of the case.

Prior to August 2018, patients undergoing lumpectomy or excisional biopsy procedures were routinely discharged with an opioid prescription (commonly acetaminophen/hydrocodone), which was usually filled at our in-house pharmacy prior to discharge. After August 30, 2018 we changed our discharge protocol to routinely prescribe the NSAID diclofenac (75 mg every 12 hours as needed) at discharge. In addition, patients were instructed to use acetaminophen over the counter (650 mg every 6 hours as needed), not exceeding 3000 mg daily. Since December 11, 2017, as part of standard postoperative care, patients are sent a “Recovery Tracker Lite” survey, in which postoperative symptoms are self-reported daily on postoperative days (PODs) 1–5 via an online patient portal. Patients report pain on the following scale: 0 = none; 1–3 = mild; 4–6 = moderate; 7–8 = severe; 9–10 = very severe. Patients completing at least one day of the post-discharge survey were included in our analysis. For patients with more than one eligible procedure during the study period, we included only the first procedure. During the first study period (December 11, 2017-August 30, 2018), opioid analgesics were routinely prescribed at discharge unless there was a medical or clinical contraindication. In the second study period (August 31, 2018-June 6, 2019) we switched to routine discharge with NSAID (diclofenac), with exceptions for patients who had medical contraindications to NSAIDs or those for whom care providers recommended discharge with an opioid analgesic.

All opioid prescriptions in our institution are placed electronically via the ePrescribe function in our electronic medical records (EMR) system. For this study, opioid medications prescribed at discharge or within 7 days of discharge were queried from the EMR, and discrepancies were reconciled by chart review. Patients were categorized as being discharged with opioids if there was documentation of a completed prescription in the EMR. In the second study period, patients who were initially discharged with NSAID were reviewed to document whether an opioid was prescribed for pain control in the 7 days following discharge.

Statistical Methods

Our goal was to assess whether there was a change in pain severity associated with cessation of routine prescription of opioids. We first evaluated secular trends, testing whether there was evidence of change over time for any outcomes during the first study period (December 11, 2017-August 30, 2018) using multivariable generalized estimating equations (GEEs) adjusting for POD of the survey (plus spline terms), age, body mass index (BMI), American Society of Anesthesiologists (ASA) score (1–2 versus 3), and operative time. We found an extremely small and non-significant decrease in the rates of moderate pain (adjusted decrease 3.6% on POD 1, p = 0.7) and a similarly small and non-significant increase in the rate of severe pain (adjusted increase 0.5% on POD 1, p = 0.8). Given that neither of these trends were statistically or clinically significant, we did not adjust for temporal trends.

We used multivariable GEEs to test the association between pain severity reported on the recovery tracker and study period. The primary endpoint was defined as moderate or higher reported pain score; therefore, a logit link function was used. Covariates included were the POD of the survey (plus spline terms to allow for nonlinearity), age, BMI, ASA score (1–2 versus 3) and operative time. The relationship between pain after discharge and intraoperative acetaminophen, intraoperative surgeon-administered local anesthetic, and the total amount of intraoperative narcotics (MMEs) has not been well established. To assess these potential confounders, we conducted a sensitivity analysis repeating the previously described analyses adding intraoperative acetaminophen, surgeon-administered local anesthetic, and intraoperative narcotics as covariates. As some patients who underwent surgery after August 30, 2018 received narcotics due to contraindications to NSAID, we repeated the primary analysis testing the association between pain severity and discharge medication (opioid versus NSAID). All analyses were performed using Stata, version 15.0 (StataCorp, College Station, TX).

RESULTS

1606 patients had lumpectomy/excisional biopsy during the study period. For 817 (49%) procedures, patients responded at least once to the recovery tracker survey pain question of interest. In patients who had more than one qualifying procedure, we studied only the first procedure, leaving 790 primary procedures and unique patients. One additional patient who received intraoperative remifentanil was excluded. 789 patients were included in our analysis.

Patient demographics differed as expected among those who responded to at least one survey pain question versus those who did not respond; patients who responded tended to be younger, have a lower ASA score and BMI, were more likely to have been treated during the second study period, and had a different racial distribution (Supplemental Table 1, p-values < 0.05). Additionally, patients who responded to at least one survey had, on average, longer operative times, required more intraoperative narcotic, and received higher doses of ketorolac (Supplemental Table 1, p-values < 0.05).

Of 789 patients who responded to at least one survey and were included in our analysis, 37% of patients responded to the pain question of interest on POD 1 and 66–68% responded on PODs 2–5. Patient characteristics were similar in both study periods and are summarized in Table 1. In the second study period, intraoperative acetaminophen use was slightly higher (97% versus 94%, p = 0.03) and post-anesthesia care unit opioid dosing was slightly lower (p = 0.014), with a smaller proportion of patients requiring > 11 MMEs (10.3% versus 5.6%). Opioid prescription at discharge decreased from 96% (315/328) in the first study period to 14% (65/461) (95% confidence interval [CI] 11%, 18%) in the second period where NSAIDs were routinely prescribed at discharge. In the second study period, 65 patients were discharged with narcotic medications for medical contraindications to NSAIDs or clinical reasons. Of 396 patients in the second study period discharged with NSAID prescription, only 5 (1.3%) were later prescribed an opioid for inadequate pain control.

TABLE 1.

Demographic and perioperative treatment characteristics, Values are reported as median (quartile 1, quartile 3) for continuous variables, and as number (percentage) for categorical variables. P-values determined by Wilcoxon Rank-sum for continuous variables, and by Fisher’s exact test for categorical variables.

Demographic variables	Study period 1 routine discharge with opioids (n = 328)	Study period 2 routine discharge with NSAID (n = 461)	p-value
Female	99%	100%	0.071
Age, years	54 (47, 66)	52 (46, 66)	0.4
BMI, kg/m2	26 (23, 30)	26 (23, 30)	0.8
ASA score
1–2	78%	77%	0.3
3	22%	23%
4	0.6%	0%
Race			0.2
African American	5.2%	6.3%
Asian	11%	14%
White	76%	72%
Other	4.6%	2.6%
Unknown	3.4%	5.4%
Perioperative variables
Anesthesia type
General	7.6%	11%	0.11
MAC	92%	89%
Operative time, minutes (IQR)	35 (29, 44)	36 (28, 45)	0.6
Intraoperative narcotic, MMEs (IQR)	10 (10, 20)	10 (10, 20)	0.8
Surgeon administered lidocaine 1%, ml (IQR)	10 (7, 10)	10 (6, 10)	0.7
Surgeon administered bupivacaine 0.5%, ml (IQR)	10 (9, 10)	10 (9, 10)	0.070
Intraoperative acetaminophen	94%	97%	0.03
Intraoperative ketorolac (mg)
0	17%	18%	0.8
15	25%	26%
30–60	59%	56%
PACU narcotics (MMEs)			0.014
0	70%	71%
1–10	20%	23%
11–30	9.1%	5.6%
31–110	1.2%	0%

NSAID non-steroidal anti-inflammatory drug, BMI body mass index, ASA American Society of Anesthesiologists, MAC monitored anesthesia care, IQR interquartile range, PACU post-anesthesia care unit, MMEs morphine milligram equivalents

The maximum pain score reported on any POD was severe in 3.8% (n = 30), moderate in 28% (n = 217), mild in 54% (n = 430), and none in 14% (n = 112). No patient reported very severe pain. On multivariable analysis adjusting for POD, age, BMI, ASA score, and operative time, the proportion of patients experiencing moderate or greater pain among those treated in the second study period compared to the first was slightly but non-significantly lower (odds ratio [OR] 0.91; 95% CI 0.67, 1.22; p = 0.5). When the analysis was adjusted for the covariates intraoperative acetaminophen, surgeon-administered local anesthetic, and intraoperative narcotics, the result was essentially unchanged (OR 0.90; 95% CI 0.66, 1.22, p = 0.5). The upper limit of the 95% CI of 1.22 excludes a clinically relevant benefit to routine opioid prescription. We also performed a comparison based on actual discharge medications (opioid versus NSAID) across both time periods, and on multivariable analysis adjusting for POD, age, BMI, ASA score, and operative time; the likelihood of experiencing moderate or greater pain was slightly but non-significantly lower among those patients discharged with NSAIDs (OR 0.87; 95% CI 0.64, 1.17; p = 0.3).

Based on the multivariable GEE model, the estimated proportion of patients experiencing moderate or severe pain decreased from PODs 1–5 (Fig. 1). We found no clinically meaningful differences in the multivariable model-estimated probability of moderate or severe pain between the two study periods (Table 2). On POD 1, the estimated risk of moderate or severe pain was 36% in the first study period compared to 34% in the second, an absolute difference of 2.3% (95% CI −4.6%, 9.1%). The upper limit of the 95% CI for routine opioid prescription is the equivalent of 22 patients who would need to be routinely sent home with opioids (number needed to treat = 1/4.6%) to prevent one occurrence of moderate or greater pain on POD 1. These estimates decreased to 7.9% versus 7.2% at POD 5, an absolute difference of 0.68% (95% CI −1.4%, 2.8%), with the 95% CI excluding an increase in pain in the second period of 1.5%.

Fig. 1.

Estimated proportion of patients reporting moderate or greater pain severity by postoperative day and period for the average patient. The prediction is generated from the multivariable GEE model adjusting for postoperative day (splines), age, BMI, ASA score (1–2 versus 3) and operative time. GEE generalized estimating equation, BMI body mass index, ASA American Society of Anesthesiologists, NSAID non-steroidal anti-inflammatory drug

TABLE 2.

Estimated probability for the outcomes of moderate pain or greater by study period and postoperative day for the average patient, adjusted for postoperative day (splines), age, BMI, ASA score (1–2 versus 3), and operative time, Absolute estimated decrease in the probability of pain going from the first study period to the second. 95% confidence intervals were calculated using the delta method

Postoperative day	Study period 1 routine discharge with opioids, % (95% CI)	Study period 2 routine discharge with NSAID, % (95% CI)	Difference, % (95% CI)
1	36 (30, 43)	34 (28, 40)	2.3 (−4.6, 9.1)
2	23 (19, 28)	22 (18, 26)	1.7 (−3.5, 6.9)
5	7.9 (5.3, 10)	7.2 (4.9, 9.4)	0.68 (−1.4, 2.8)

CI confidence interval, NSAID non-steroidal anti-inflammatory drug

DISCUSSION

Although patients undergoing ambulatory procedures are routinely discharged with opioid prescriptions, we were able to exclude a clinically relevant benefit for routine discharge with opioids. Our results were similar when analyzed by study period or by discharge with opioid versus NSAID prescription.

Consistent with our findings, prior studies in the ambulatory surgery setting have shown that acute post-surgical pain can be effectively managed with non-opioid medications such as NSAIDs or acetaminophen.^14–19 In an open randomized trial, Stessel et al. evaluated the efficacy of paracetamol/naproxen (an NSAID) compared to paracetamol/controlled release (CR) oxycodone in treating acute postoperative pain after various ambulatory surgery procedures.¹⁴ In contrast to our study, the authors noted that at their institution, paracetamol/naproxen was the gold standard for treating postoperative pain in patients undergoing ambulatory procedures, and hypothesized that patients treated with paracetamol/oxycodone would achieve better pain relief with movement. 105 patients undergoing knee arthroscopy or inguinal hernia repair were randomized to receive paracetamol/naproxen or paracetamol/CR oxycodone for 24 hours, or paracetamol/CR oxycodone for 48 hours. Investigators found no significant difference in pain with movement or at rest among patients in the paracetamol/naproxen group compared to paracetamol/CR oxycodone for 24 hours (p = 0.6) or paracetamol/CR oxycodone for 48 hours (p = 0.7). Similarly, Raeder et al. performed a randomized double-blinded study of 800 mg ibuprofen compared to codeine/acetaminophen for pain control in the first 72 hours after ambulatory surgery.¹⁷ In 104 patients undergoing elective hernia repair, hemorrhoidectomy, or varicose vein resection, results demonstrated no significant difference in pain control between the two groups. Although breast procedures were not included in these studies, results demonstrate that postoperative pain in a variety of ambulatory surgery cases can be adequately managed with alternatives to opioids.

At least two other studies have examined the efficacy of a non-opioid pain regimen in the ambulatory breast surgery setting. In a Canadian study, Mitchell et al. performed a randomized double-blinded trial examining the equivalence of ibuprofen plus acetaminophen compared to codeine/acetaminophen/caffeine in patients undergoing breast surgery procedures.¹⁸ Of 141 breast cancer patients enrolled in the study, approximately 80% had lumpectomy with or without an axillary procedure, and 20% underwent mastectomy. There was no difference between the two groups in terms of reported pain intensity (p = 0.8) or pain relief (p = 0.5) over 7 postoperative days. This study demonstrated the equivalence of an NSAID/acetaminophen regimen to an opioid/acetaminophen regimen for post-discharge pain control in a population of patients which, compared to those in our study, had more complex and presumably more painful breast procedures. Our study included patients undergoing lumpectomy or excisional biopsy without an axillary procedure, and, as expected, most patients (68%) reported no pain or mild pain. Given the possible adverse effects of opioids, the potential for abuse, and the low pain severity of lumpectomy/excisional biopsy procedures, it is reasonable to consider discharge with non-opioid analgesics. Rojas et al. reported on their experience with a non-opioid discharge protocol in a pilot study including patients with and without breast cancer undergoing lumpectomy.²⁰ The authors utilized an ERAS protocol including preoperative gabapentin, intraoperative ketorolac, and intraoperative injection of liposomal bupivacaine with patients discharged on ibuprofen/acetaminophen only. Patients on the ERAS protocol were compared to those undergoing usual care who were discharged with a median of 54 MMEs (range 0–120), a seemingly high amount of opioid for this procedure. Investigators found no difference in reported POD 1 and POD 7 pain scores between the two groups. In contrast to our practice, their perioperative protocol included administration of gabapentin and injection of liposomal bupivacaine which, based on our results, appear to be unnecessary in patients undergoing lumpectomy or excisional biopsy alone. Despite differences in perioperative management, these studies all demonstrate the feasibility of discharge with a non-opioid pain control regimen and are consistent with our results. In a population of patients undergoing lumpectomy or excisional biopsy, a switch to routine NSAID prescription achieved pain control comparable to an opioid regimen, and supports the use an NSAID in lieu of opioids for those patients without an NSAID contraindication.

This study has limitations, as it was based on patient-reported pain scores from survey respondents. Survey respondents tended to be younger and healthier, possibly indicating a bias toward underestimating pain in this cohort. On the other hand, respondents also tended to have indicators of more difficult surgeries, which could also reflect a bias toward overestimating pain. It is difficult to assess how these differences in characteristics of responders versus nonresponders might impact the estimated rates of moderate or greater pain. However, we have no reason to suspect that there would be a difference in the responder bias by study period. Therefore, although our population of respondents could have been biased, we are confident that our primary result, the absolute risk difference by study period, is accurate. Additionally, we were unable to assess how many of the opioid or NSAID tablets that were prescribed in both time periods were taken by the patient, as these data were not collected, and so our results are based on prescriptions rather than actual medication usage. We do not expect that the relationship between the number of pills prescribed and the number of pills taken would differ by study period. Based on existing literature demonstrating that in patients undergoing general surgery procedures, up to 70% will not use the opioids they are prescribed²¹, it is also likely that there was low utilization of opioids in the first study period. Patients who may have obtained opioids from a provider outside Memorial Sloan Kettering Cancer Center following their surgical procedure were not captured in this study, as we utilized our institutional EMR. Lastly, the estimated risk of experiencing moderate or severe pain was > 30% on POD 1 in both study periods, suggesting that further investigation into effective pain management interventions in the early postoperative period are necessary. A strength of our study is that the likelihood of it being subject to confounding is low, given that we were studying a very standard operation and postoperative pathway over a relatively short period of time. We did not see any evidence of secular trends, and patient characteristics were very similar during the two periods. Results for the analysis comparing routine discharge orders for opioids versus NSAIDs demonstrated very similar estimates of pain reduction in the second period (OR 0.91) and for discharge with NSAIDs (OR 0.87). Hence, we believe that our central estimate (OR 0.91) is a good estimate of the causal effect of avoiding opioid prescription.

Conclusions

In patients undergoing lumpectomy and excisional biopsy procedures, we can exclude a clinically meaningful difference in reported post-discharge pain scores with the routine use of opioids versus NSAIDs at discharge. Patients undergoing these procedures should not routinely be discharged with opioids. Further studies supporting the feasibility and safety of opioid minimization or elimination during the postoperative period for other surgical procedures are warranted.

Synopsis:

Here we compare patient-reported post-discharge pain scores among patients treated before/after the change from routine discharge opioid prescription to an NSAID for patients having lump/ex. We conclude that patients undergoing lump/ex should not routinely be discharged with opioids.