NSAID “allergy” labeling is associated with increased postpartum opioid utilization

Abstract

Background:

Current guidelines recommend a stepwise approach to postpartum pain management, beginning with acetaminophen and nonsteroidal anti-inflammatory drugs (NSAIDs) and with opioids added only if needed. Report of a prior NSAID-induced adverse drug reaction (ADR) may preclude use of first-line analgesics, despite evidence that many patients with this allergy label may safely tolerate NSAIDs.

Objective:

To assess the association between reported NSAID ADRs and postpartum opioid utilization.

Methods:

We performed a retrospective cohort study of birthing people who delivered within an integrated health system (1/1/2017–12/31/2020). Study outcomes were postpartum inpatient opioid administrations and opioid prescriptions at discharge. Statistical analysis was performed on a propensity score-matched sample, which was generated with the goal of matching to the covariate distributions from individuals with NSAID ADRs.

Results:

Of 38,927 eligible participants, there were 883 (2.3%) with an NSAID ADR. Among individuals with reported NSAID ADRs, 49.5% received inpatient opioids in the postpartum period, compared with 34.5% of those with no NSAID ADRs (difference=15.0%, 95% CI [11.4–18.6%]). For patients who received postpartum inpatient opioids, those with NSAID ADRs received a higher total cumulative dose between delivery and hospital discharge (median 30.0 vs 22.5 morphine milligram equivalents [MME] for vaginal deliveries; median 104.4 vs 75.0 MME for cesarean deliveries). The overall proportion of patients receiving an opioid prescription at the time of hospital discharge was higher for patients with NSAID ADRs compared to patients with no NSAID ADRs (39.3% vs 27.2%, difference=12.1%, 95% CI [8.6–15.6%]).

Conclusion:

Patients with reported NSAID ADRs had higher postpartum inpatient opioid utilization and more frequently received opioid prescriptions at hospital discharge compared to those without NSAID ADRs, regardless of mode of delivery.

Capsule Summary:

In this cohort study, presence of NSAID ADR labeling was associated with increased inpatient opioid utilization in the immediate postpartum period, and higher rates of opioid prescribing at hospital discharge following delivery.

INTRODUCTION

Over 3.6 million births occur annually in the United States, and postpartum pain is common (1). Effective pain management is crucial as acute postpartum pain has been linked to adverse outcomes, including increased risk of postpartum depression and chronic pain (2, 3).

Current national guidelines recommend a stepwise multimodal approach for pain control after vaginal or cesarean delivery, starting with first-line nonopioid analgesics such as acetaminophen and nonsteroidal anti-inflammatory drugs (NSAIDs), followed by lower and then higher potency opioids only as needed (4). However, despite these recommendations, both inpatient opioid consumption and outpatient opioid prescribing after childbirth remains prevalent (5–7). Exposure to opioids in the postpartum period is not only associated with immediate opioid-related side effects such as nausea, drowsiness and constipation, but also with sustained opioid use and heightened risk for serious opioid-related events (4, 8–11).

Studies have demonstrated that peripartum use of NSAIDs can effectively lower patient pain scores and decrease opioid use and sedation (12). For some patients, however, NSAIDs may be contraindicated due to report of a prior NSAID-induced adverse drug reaction (ADR), which are reported by about 2% of women of childbearing age (13). The majority of reported NSAID ADRs captured in the medical record are thought to represent nonallergic intolerances or side effects as assessed by clinical history alone (14). For the minority of patients with history of an NSAID-induced reaction that could be consistent with true hypersensitivity, evaluation by an allergy specialist allows for clearance of NSAID use over 85% of the time (15). However, drug allergy labels often remain unverified (16) and inaccurate NSAID ADR reporting may influence clinical decision making and prescribing practices. Our previous studies have demonstrated associations between reported NSAID ADRs and increased opioid prescribing in patients with chronic pain and following orthopedic surgery (17, 18).

Given the high number of deliveries that take place annually for which postpartum pain management is required, understanding the extent to which unverified NSAID ADR reporting affects postpartum pain management and opioid prescribing is vital for identifying targetable risk factors and novel methods to allow use of first-line multimodal analgesic approaches and reduce unnecessary opioid exposures for a substantial number of patients. In this study, we sought to examine opioid utilization for individuals with and without reported NSAID ADRs in the postpartum period.

METHODS

Data Source

We conducted a retrospective cohort study of birthing people who underwent vaginal or cesarean delivery at Mass General Brigham (MGB), a single integrated healthcare system that includes specialty and community hospitals in Massachusetts. Patient clinical information, including sociodemographic, allergy, and medication prescription and administration data, were obtained from the MGB Research Patient Data Registry and Enterprise Data Warehouse databases, which house data for over four million individuals receiving care at MGB (19, 20).

Study Cohort and Follow-up

Patients were included in the cohort if they were 1) ≥ 18 years old at time of childbirth, and 2) had a delivery between 1/1/2017–12/ 31/2020. Deliveries were identified by Current Procedural Terminology codes (6, 11). If a patient had multiple deliveries during the study period, only the first one was included.

We excluded patients with a prolonged hospital length of stay of >30 days, a diagnosis of opioid use disorder as defined by presence of two or more International Classification of Diseases, Ninth Revision (ICD-9) or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes (18) prior to the delivery date, or those who received methadone or buprenorphine during the delivery admission to minimize confounding. We also excluded individuals with a malignant cancer diagnosis in the one year before or after the delivery date, as malignancy can influence opioid use (21).

Exposure and Covariate Identification

The primary exposure was presence of any reported ADR to aspirin or another NSAID documented in the medical record and listed as active at the time of delivery. Patients with their first NSAID ADR documented post-delivery were excluded to establish temporality (e.g., documentation of NSAID ADR before postpartum opioid prescribing). Characteristics of reported NSAID ADRs were examined.

We measured covariates in the period prior to delivery using coded diagnoses available in the electronic health record. Characteristics of interest included age, self-reported race and ethnicity, insurance information, mode of delivery, and pain-associated medical comorbidities as defined by presence of at least 2 ICD-9 or ICD-10 diagnoses codes documented prior to delivery (14, 17, 22, 23). Other covariates that may be related to opioid use were also obtained and included healthcare utilization (inpatient, outpatient, and emergency department encounters) and presence of opioid prescriptions in the medical record in the year prior to delivery, peripartum obstetric complications (24, 25), and hospital length of stay.

Outcome Assessment

We identified several measures of inpatient opioid administrations in the postpartum period, defined as from the time of delivery to hospital discharge. Opioids of interest included: codeine, fentanyl, hydrocodone, hydromorphone, meperidine, morphine, oxycodone, oxymorphone, and tramadol (22, 26, 27). We evaluated for presence of intrapartum opioid administrations documented as given through neuraxial (e.g., epidural, spinal, or combined spinal-epidural) or patient-controlled analgesia (PCA) routes. Total doses of intrapartum opioids administered through these routes are not fully extractable from the electronic health record and therefore were not quantified.

We assessed for presence of postpartum opioid administrations which included oral, intravenous, sublingual, or transdermal routes of delivery (e.g., administrations exclusive of neuraxial or PCA-administered routes). Cumulative postpartum opioid doses were calculated and converted into morphine milligram equivalents (MME) for standardized units of comparison across opioids (22, 28). As opioid utilization can also affect gastrointestinal mobility, we examined postpartum inpatient laxative (docusate, bisacodyl, senna, polyethylene glycol, and lactulose) administrations.

We obtained opioid medication prescription data at the time of hospital discharge. We performed a validation for both inpatient opioid administrations and outpatient opioid prescriptions by randomly selecting 50 patients for manual review, and all individuals classified as receiving an opioid administration or prescription were correctly identified.

Statistical Analysis

Descriptive statistics were calculated for all variables, including rates and cumulative doses of opioids administered postpartum. Continuous variables are reported as means with standard deviations (SDs) or medians with interquartiles and compared using Student’s t-tests or Wilcoxon rank-sum tests, as indicated by the distribution shape of the data. Categorical variables are reported as counts with percentages and compared using chi-square tests. We performed stratified analyses to examine the association of NSAID ADRs with postpartum opioid utilization by mode of delivery (vaginal or cesarean).

We used propensity score-matching to address the potential for confounding that can occur in any observational study. First, we built a logistic regression model using measured covariates (age, race/ethnicity, insurance, delivery type, hospital, year of delivery, length of hospital stay from delivery to discharge, prior healthcare utilization, presence of opioid prescription prior to delivery, peripartum complications, medical comorbidities) to generate propensity scores as indicators of propensity to have a reported NSAID ADR. Next, we matched patients with reported NSAID ADRs to those without reported NSAID ADRs in a 1:5 ratio based on propensity scores. Covariate balance between propensity score-matched groups was assessed and a standardized mean difference of < 0.1 between groups was considered well matched (29).

The associations between reported NSAID ADRs and inpatient opioid utilization and outpatient opioid prescribing were assessed using chi-square tests for proportions and Wilcoxon rank-sum tests for cumulative doses. The frequency of laxative administration was compared using Wilcoxon rank-sum tests. Since the two groups in the propensity score were well matched on all observable variables, we did not perform an additional adjustment for covariates. Analyses were performed using SAS, version 9.4, between 1/2023–5/2023. A 2-sided 5% significance level (P<0.05) was set for all statistical tests. This study was approved by the MGB Institutional Review Board. The STROBE guidelines were used to ensure the reporting of this observational cohort study (30).

RESULTS

A total of 38,927 birthing people, including 24,988 (64.2%) with vaginal delivery and 13,939 (35.8%) with cesarean delivery, met eligibility criteria for inclusion in the study cohort (Figure 1). Of these, 883 (2.3%) patients had an active NSAID ADR listed in the medical record prior to delivery.

Figure 1.

Flowchart of patient inclusions and exclusions. Only the first delivery was included for patients with more than one delivery during the study period.

The mean age of participants in the entire cohort was 32.8 years (SD 5.1), with 22,813 (58.6%) individuals who identified as non-Hispanic white, 3,998 (10.3%) as non-Hispanic black, and 2,951 (7.6%) as Hispanic. There were some differences in baseline characteristics between patients with and without NSAID ADRs in the entire cohort (Table 1). The 1:5 propensity score-matched sample included a total of 5,298 participants. Measured covariates and baseline comorbidities were well-balanced after matching for all patients combined (Table 1) and when stratified by route of delivery (Table E1 in the Online Repository). All subsequent analyses were conducted using the propensity score-matched sample.

Table 1.

Characteristics of patients who underwent vaginal or cesarean delivery with or without NSAID adverse drug reactions (ADRs)

		Comparison group from the entire cohort	Comparison group from 1:5 propensity score-matched sample
	NSAID ADR, n=883	No NSAID ADR, n=38,044	No NSAID ADR, n=4,415	Standardized Difference (no ADR vs ADR)b
Age, mean years [SD]	33.3 [5.4]	32.8 [5.1]	33.4 [5.3]	−0.01
Race and Ethnicity, count (%)
Hispanic	111 (12.6)	2,840 (7.5)	521 (11.8)	−0.03
Non-Hispanic Asian	66 (7.5)	4,269 (11.2)	318 (7.2)	−0.01
Non-Hispanic Black	135 (15.3)	3,863 (10.2)	676 (15.3)	0.00
Non-Hispanic White	440 (49.8)	22,373 (58.8)	2,216 (50.2)	0.01
Othera	101 (11.4)	3,662 (9.6)	515 (11.7)	0.01
Unknown	30 (3.4)	1037 (2.7)	169 (3.8)	--
Insurance, count (%)
Public	318 (36.0)	9,738 (25.6)	1,602 (36.3)	0.01
Private	557 (63.1)	27,851 (73.2)	2,782 (63.0)	0.00
Other/Self-Pay/None	8 (0.9)	455 (1.2)	31 (0.7)	--
Delivery year, count (%)
2017	238 (27.0)	10,705 (28.1)	1,204 (27.3)	0.01
2018	256 (29.0)	10,172 (26.7)	1,297 (29.4)	0.01
2019	209 (23.7)	9,098 (23.9)	1,006 (22.8)	−0.02
2020	180 (20.4)	8,069 (21.2)	908 (20.6)	--
Delivery Type, count (%)
Vaginal	530 (60.0)	24,458 (64.3)	2,650 (60.0)	0.0
Cesarean	353 (40.0)	13,586 (35.7)	1,765 (40.0)
Total hospital length of stay from admission to discharge, mean days [SD]	3.8 [2.4]	3.5 [2.0]	3.8 [2.4]	0.01
Hospital length of stay from delivery to discharge, mean days [SD]	2.9 [1.3]	2.8 [1.1]	3.0 [1.4]	0.00
Complication during hospital admission, count (%)	193 (21.9)	7,436 (19.5)	976 (22.1)	0.01
Hospital encounters in the prior 1 year, mean [SD]
Inpatient admissions	0.8 [1.0]	0.7 [0.9]	0.7 [1.1]	0.03
Outpatient encounters	9.3 [6.9]	7.7 [5.9]	9.3 [6.8]	0.01
Emergency room encounters	0.7 [1.6]	0.4 [0.9]	0.6 [1.5]	0.03
Opioid prescription within 1 year before delivery, count (%)	85 (9.6)	2,103 (5.5)	430 (9.7)	0.00
Medical co-morbidities, count (%)
Diabetes	20 (2.3)	373 (1.0)	94 (2.1)	−0.01
Chronic back pain	51 (5.8)	919 (2.4)	255 (5.8)	0.00
Depression	69 (7.8)	1,650 (4.3)	352 (8.0)	0.01
Anxiety	84 (9.5)	2,692 (7.1)	426 (9.6)	0.00
Fibromyalgia	8 (0.9)	51 (0.1)	34 (0.8)	−0.02
Rheumatoid arthritis	12 (1.4)	81 (0.2)	54 (1.2)	−0.01

^aAmerican Indian/Eskimo, Native Hawaiian/Pacific Islander, Multiple races, Other

^bExact match for delivery type; standardized difference value of “0.00” indicates difference <0.005, and “--” denotes the reference subgroup for categorical variables

Abbreviations: ADR, adverse drug reaction; NSAID, non-steroidal anti-inflammatory drug; SD, standard deviation; IQR, interquartile range

Nearly all patients received intrapartum opioid medications (epidural, spinal, or PCA routes) during labor and delivery, with no differences in utilization between those with and without NSAID ADRs (87.1% vs 88.0% among all deliveries combined; stratified: 81.1% vs 82.3% for vaginal deliveries, 96.0% vs 96.5% for cesarean deliveries).

Frequency of inpatient postpartum opioid use

Among individuals with reported NSAID ADRs, 49.5% received opioids in the postpartum period, compared with 34.5% of patients with no NSAID ADRs (difference=15.0%, 95% CI [11.4–18.6%]; p<0.001). For those with vaginal deliveries, 27.7% vs 13.4% of patients with and without NSAID ADRs, respectively, received inpatient postpartum opioids (difference=14.3%, 95% CI [10.3–18.4%]; p<0.001); for those with cesarean deliveries, 82.2% vs 66.2% of individuals with and without NSAID ADRs received inpatient postpartum opioids (difference=16.0%, 95% CI [11.4–20.5%]; p<0.001). While there was a decrease in overall proportion of individuals receiving inpatient postpartum opioids over time, those with NSAID ADRs more frequently received opioids regardless of delivery route or year of delivery (Figure 2). Among patients who received inpatient postpartum opioids, oxycodone was the most frequently administered opioid for patients with and without NSAID ADRs.

Figure 2.

Proportion of patients receiving inpatient postpartum opioids from time of delivery to hospital discharge in the 1:5 propensity score-matched sample (p-values indicated).

Inpatient postpartum opioid cumulative doses

The cumulative inpatient opioid dose utilized by patients who received postpartum opioids is shown in Figure 3, stratified by route of delivery. Combining across all years, the median cumulative opioid dose received for birthing people with and without NSAID ADRs was 30.0 (IQR 15.0–53.3) versus 22.5 (IQR 7.5–45.0) MME among those with vaginal deliveries, and 104.4 (IQR 59.5–172.5) versus 75.0 (IQR 30–127.5) MME among those with cesarean deliveries. Cumulative doses of inpatient postpartum opioids administered were similar for individuals with vaginal delivery regardless of NSAID ADR status when stratified by year. For patients undergoing cesarean delivery, those with NSAID ADRs consistently received higher cumulative postpartum opioid doses when stratified by year, despite a decrease in overall median doses administered over time. Patients with cesarean deliveries and NSAID ADRs also received more laxatives in the postpartum period compared to those with no NSAID ADRs (median 9.0 [IQR 7.0–11.0] vs 8.0 [6.0–11.0] administrations, p=0.046) (Table E2 in the Online Repository).

Figure 3.

Cumulative inpatient opioid dose administered from time of delivery to hospital discharge among patients with A) vaginal deliveries or B) cesarean deliveries who received post-partum inpatient opioids in the 1:5 propensity score-matched sample. Doses measured in morphine milligram equivalent (median, interquartile range), and p-values are indicated.

Opioid prescribing at hospital discharge

The overall proportion of patients who received an opioid prescription at hospital discharge following childbirth was higher for those with reported NSAID ADRs (39.3%) compared to no NSAID ADRs (27.2%) (difference=12.1%, 95% CI [8.6–15.6%]), and this pattern persisted across delivery types (14.7% vs 4.9% for those with vaginal deliveries, difference=9.9%, 95% CI [6.7–13.0%]; 76.2% vs 60.7% for those with cesarean deliveries, difference=15.5%, 95% CI [10.5–20.5%]). Notably, there were decreases over time from 2017 to 2020 in the proportion of individuals prescribed an opioid at time of hospital discharge for patients with both vaginal and cesarean deliveries (Figure 4). However, the association between presence of NSAID ADR and higher rates of opioid prescriptions at discharge remained even after stratification by route of delivery and year of delivery. There were no differences in median cumulative dose of opioids prescribed at time of discharge by NSAID ADR status (Figure E1 in the Online Repository), and oxycodone was the most frequently prescribed opioid for all participants.

Figure 4.

Proportion of patients receiving an opioid prescription at hospital discharge in the 1:5 propensity score-matched sample (p-values indicated).

NSAID ADR characteristics

A total of 1,071 NSAID ADR entries were reported (Table E3 in the Online Repository). Based on reaction descriptions associated with NSAID ADRs captured in the medical record, 59.5% included symptoms that could be consistent with hypersensitivity: angioedema, urticaria, rash, bronchospasm or wheeze, anaphylaxis, hypotension, or itching. There were 200 entries (18.7%) with reaction characteristics suggestive of medication side effects or intolerances, and an additional 257 (24.0%) entries with an “unknown” or “other” reaction listed.

DISCUSSION

The results of this cohort study indicate that NSAID ADR labeling was associated with higher rates of inpatient postpartum opioid utilization and larger cumulative opioid doses received in the inpatient postpartum setting. This relationship remained after stratification by delivery type. The observed differences were larger for patients with cesarean deliveries and persisted even after stratification by year of delivery despite an overall decrease in inpatient postpartum utilization over time. At the time of hospital discharge, we found an overall 12.1-percentage point higher opioid prescription rate among individuals with, compared to those without, reported NSAID ADRs (9.8-percentage point and 15.5-percentage point higher for patients with vaginal and cesarean deliveries, respectively).

Postoperative opioids have been associated with persistent opioid use in patients undergoing minor and major surgeries (22, 31, 32), and recent studies indicate opioid exposure after childbirth to be similarly associated with increased risk for chronic use regardless of mode of delivery (9, 10, 33). In the US, 98.4% of births occur in a hospital setting and childbirth is the most common cause of hospitalization (34, 35). We observed both higher frequencies of postpartum opioid inpatient utilization and discharge opioid prescriptions for patients with NSAID ADRs in this study. Notably, pregnancy-associated mortality involving opioids substantially rose between 2007–2016 (36) and rates of drug overdose mortality increased over 80% from 2017–2020 among pregnant and postpartum persons (37). Taken together, excessive opioid use unnecessarily exposes patients to medications with addictive potential, and overprescribing may increase collateral risk related to unused opioid tablets and potential for exposure, misuse, or diversion (38, 39). Moreover, opioid-associated side effects including nausea, sedation and constipation are common (4), and we found that individuals with NSAID ADRs and cesarean delivery received more laxatives in the postpartum period prior to hospital discharge. It is imperative for clinicians to carefully consider whether an opioid is required, or if an alternative nonopioid analgesic may be safely used.

Stepwise multimodal approaches beginning with acetaminophen and NSAIDs with escalation to opioids only if needed have been shown to decrease pain scores, while also reducing inpatient opioid exposures and frequency of discharge opioid prescriptions (12, 40, 41). Furthermore, elimination of opioids from routine post-partum pain management has not been associated with changes in patient satisfaction with pain relief (42). This approach is now guideline-recommended for pain management following both vaginal and cesarean delivery (4). A cohort study of 2,503 postpartum patients who delivered at a tertiary care center in the Midwest observed an overall decrease in the frequency and amount of postpartum opioid use from 2017–2019, likely attributable to increased awareness and changes in opioid prescribing practices over time (7, 43). We similarly found that rates of postpartum inpatient opioid utilization and discharge opioid prescriptions in our cohort decreased from 2017 to 2020, although those with NSAID ADRs consistently had more opioid exposures compared to patients with no NSAID ADRs even when stratified by year. Future work should explore other barriers to implementation of first-line pain management strategies to promote analgesic stewardship.

Prior studies have estimated that <20% of NSAID ADR entries captured in the electronic health record are consistent with hypersensitivity by clinical history (14). In contrast, nearly 60% of NSAID ADR entries included reaction characteristics suggestive of hypersensitivity in our cohort of individuals of childbearing age, which may be due to differences in the population studied. Only 18.7% of ADRs in our study were consistent by history with side effects or intolerances, which typically do not represent absolute contraindications for future NSAID use, and 24% represented reactions with an unclarified or unknown reaction history. These findings highlight the particular important role of allergists in the assessment and management of patients of childbearing age with unverified NSAID ADRs.

Notably, in a cohort of patients referred to allergy specialists for evaluation of NSAID hypersensitivity and who underwent diagnostic testing, >85% of supervised NSAID drug challenges were negative for an immediate or delayed reaction and allowed for use of a clinically indicated NSAID (15). If applied to the findings in this study, where the absolute risk reduction of receiving an inpatient postpartum opioid was 15%, we estimate that allergist evaluation with diagnostic testing for approximately 8 patients with reported NSAID allergy would allow for prevention of 1 individual from receiving opioids in the inpatient postpartum period. With over 3.6 million births in 2022 (44), this estimate extrapolates to potential avoidance of opioids for more than 10,500 patients.

While pregnancy has previously been considered a contraindication for allergy drug testing, recent studies have demonstrated the safety of antibiotic allergy evaluation and testing performed in pregnancy (45, 46). A recent case series described NSAID challenges performed during pregnancy for four women requiring aspirin for pre-eclampsia (47). All challenges were negative which allowed for allergy delabeling and clearance for use of both aspirin and other NSAIDs. Recent updates to national drug allergy practice parameters now recommend graded drug challenge for patients with acute coronary syndromes and urgent need for aspirin, or patients requiring NSAIDs for pain; however, the special situation of pregnancy is not described (46). In light of evidence demonstrating the safety of drug allergy testing in pregnancy and important real-world downstream consequences of NSAID allergy labeling, we suggest that graded medication challenges should be strongly considered for appropriate candidates regardless of pregnancy status. Quality improvement initiatives involving multiple specialties including obstetrics, family medicine and anesthesiology should additionally consider specifically targeting this population to address unconfirmed NSAID ADRs as these medications may be indicated for multiple reasons during pregnancy and the postpartum period. Future work to assess the impact and outcomes following allergy specialist intervention is also warranted.

Limitations

This study is subject to several limitations. First, our study included participants from a single tertiary healthcare system in the Northeastern United States and therefore the results may not be generalizable to patients receiving care at other care centers or geographic locations. However, others have studied regional differences in opioid dispensation and found that patients in the Northeast are less likely to receive opioids at discharge following delivery compared to other areas in the US (6). Second, we were unable to quantify neuraxial and PCA-administered opioids during the delivery hospitalization due to limitations in data extraction from the electronic health record. Therefore, the cumulative dose of postpartum opioids reported is exclusive of opioids administered via the aforementioned routes. However, this study’s use of electronic health record-based datasets did allow for collection and analysis of detailed patient medical information and medication utilization data which are often not available through national and claims-based sources. Thirdly, while missing covariate information is possible due to patients seeking care outside of our healthcare system, a thorough assessment of each patient’s medical history and comorbidity information is standard practice at our institution and captured during both outpatient clinical visits and at the time of hospital admission. A sensitivity analysis limited to patients with one or more primary care visits at MGB in the year prior to delivery was also performed, and results were similar to our main study findings. Finally, our prescription database files include clinician prescribed medications orders, but are not linked to pharmacy fill data. As such, we are unable to determine what percentage of discharge opioid prescriptions were subsequently filled and dispensed.

CONCLUSIONS

This retrospective cohort study found that reported NSAID ADRs were associated with significantly higher opioid utilization for patients in the postpartum period. In particular, we found increases in the frequency of individuals receiving inpatient opioids, cumulative doses of postpartum opioids administered, and proportions of opioids prescribed at hospital discharge. Morbidity and mortality associated with opioid use is widespread. As greater than 3 million births occur in the US annually, identification of targetable risk factors associated with opioid use is critical to help curb unnecessary prescribing at the time of childbirth. Importantly, NSAID ADR labeling frequently does not equate to true hypersensitivity, and multi-disciplinary care of obstetric patients involving allergy specialists may provide opportunities for potential allergy delabeling and allow for use of first-line opioid-sparing pain management strategies in the postpartum period.

Clinical Implications:

NSAID ADR labels are associated with increased opioid utilization, although they do not necessarily equate to true NSAID hypersensitivity. Involving allergists in pregnancy care pre-delivery may allow for allergy delabeling and reduce unnecessary opioid use.

Abbreviations:

ADR

adverse drug reaction

CI

confidence interval

COX

cyclooxygenase

EHR

electronic health record

ICD

International Classification of Diseases

NSAID

non-steroidal anti-inflammatory drug

OR

odds ratio

PS

propensity score

SD

standard deviation

MGB

Mass General Brigham