Multimodal Analgesia’s Impact on Opioid Use and Adverse Drug Effects in a Multihospital Health System

Abstract

Purpose:In addition to opioid abuse and dependency, opioid use can lead to opioid related adverse drug events (ORADEs). ORADEs are associated with increased length of stay, cost of care, 30-day readmission rate, and inpatient mortality. The addition of scheduled non-opioid analgesic medications has shown to be effective in reducing opioid utilization in post-surgical and trauma populations, but evidence in entire hospital patient populations is limited. The objective of this study was to determine the effects of a multimodal analgesia order set on opioid utilization and adverse drug events in adult hospitalized patients. Methods: This retrospective pre/post implementation analysis was conducted at 3 community hospitals and a level II trauma center between January 2016 and December 2019. Patients included were 18 years of age or older, admitted for greater than 24 hours, and had at least one opioid ordered during hospital admission. The primary outcome of this analysis was the average oral morphine milligram equivalents (MME) used on days 1 through 5 of hospitalization. Secondary outcomes included the percentage of hospitalized patients with an opioid ordered for analgesia who received a scheduled non-opioid analgesic medication, the average number of ORADEs recorded in nursing assessments on hospitalization days 1 through 5, length of stay, and mortality. Multimodal analgesic medications included acetaminophen, gabapentinoids, non-steroidal anti-inflammatory drugs, muscle relaxants, and transdermal lidocaine. Results: The pre- and post-groups included 86 535 patients and 85 194 patients, respectively. The average oral MMEs used on days 1 through 5 were lower in the post-group (P < .0001). Utilization of multimodal analgesia as measured by the percentage of patients with 1 or more scheduled multimodal analgesia agent ordered increased from 33% to 49% at the end of the analysis. Conclusion: Utilization of a multimodal analgesia order set was associated with a decrease in opioid use and an increase in multimodal analgesia use in an entire hospital adult population.

Introduction

Balancing safe opioid prescribing while providing adequate analgesia continues to be a complex challenge for healthcare providers. In 2019, there were nearly 50 000 drug overdose deaths in the United States involving prescription opioids, heroin, and illicitly manufactured fentanyl. ¹ Cicero and colleagues analyzed data from a survey of 2797 patients seeking treatment for heroin use or dependency and found that 75% of those who’s addiction began in the 2000s reported that their first regular opioid used was a prescription medication. ² The total economic burden of prescription opioid overdose, abuse, and dependence in the United States was estimated to be 78.5 billion dollars in 2013. ³ As the opioid epidemic continues, the Society of Hospital Medicine released a statement on the management of acute non-cancer pain in hospitalized adults emphasizing the consideration of non-opioid analgesia prior to use of opioids. ⁴ Furthermore, the 2019 Clinical Guidelines for Pain Management in Acute Musculoskeletal Injury recommends the use of multimodal analgesia versus monotherapy with opioids and the tailoring of multimodal analgesia based on patient injuries and conditions. ⁵

In addition to abuse and dependency concerns, opioid related adverse drug events (ORADEs) are increasingly recognized as a clinical and economic concern. ORADEs are associated with increased length of stay, cost of care, 30-day readmission rate, and inpatient mortality in multiple studies.^{6 -8} Patients who experienced ORADEs while in a hospitalized setting were more likely to have received a higher total dose of opioids during hospitalization. ⁶ ORADEs were prevalent in 10.6% of patients, associated with a nearly 3% increase in absolute mortality, and a $8225 average increase in index hospitalization cost. ⁶ This study seeks to determine the effects of a multimodal analgesia order set on opioid utilization and adverse drug events in adult hospitalized patients.

Materials and Methods

Patients and Intervention

This retrospective, multicenter analysis took place in a health-system of 3 community hospitals and a non-teaching level II trauma tertiary center, totaling 886 beds. This study was approved by the health-system’s Ethics Committee. Two groups were identified before and after implementation of a multimodal analgesia order set in January 2018. Cohorts consisted of a pre-group from January 1st, 2016 to December 31st, 2017 and a post-group from January 1st, 2018 to December 31st, 2019. Inclusion criteria consisted of patients aged 18 years or older, with a hospital length of stay greater than 24 hours, and a scheduled or “as needed” opioid ordered for pain control during the hospital admission.

The order set implemented consisted of preselected multimodal analgesia options, including acetaminophen, gabapentinoids, non-steroidal anti-inflammatory drugs (NSAIDs), muscle relaxants and transdermal lidocaine (Figure 1). The order set also guided clinicians to select appropriate doses of opioid analgesics based on the patient’s history of opioid use. The order set was made available during the hospital admission as a standalone order set. In addition, the multimodal analgesia order set was embedded into the general admission order set, which helped increase its utilization. Prescribers were not required to use the order set and each component remained available as single-entry items. Awareness and education of the multimodal order set was promoted through in-services at physician committee meetings as well as informational sheets posted at physician workstations.

Figure 1.

Example of a multimodal analgesia order set.

Data Collection

Demographic data, principal diagnosis, and history of opioid use, dependency, or toxicity were collected from Midas+ Care Management™. Opioid use or dependency was determined using the International Classification of Diseases 10th Revision (ICD-10) codes F11.1, F11.2, and F11.9. Opioid toxicity was determined by the presence of ICD-10 codes T40.0 through T40.6. Corresponding ICD-10 codes were also used to determine the presence of chronic pain, diabetes mellitus, heart failure, and chronic obstructive pulmonary disease (COPD). The 5 most common pain locations and 10 most common surgical procedures were determined using ICD-10 codes. Surgical procedures in the first 5 days of hospital stay were determined by operating room records. Oral MMEs administered and ORADEs were collected using custom reports in Allscripts Clinical Performance Management™. ORADEs consisted of possible opioid related symptoms documented by nursing staff, including nausea, vomiting, constipation, confusion, and bradypnea. Length of stay and mortality were obtained from Midas+ Care Management™. The percentage of multimodal use was determined through a denominator of individual patients with opioid orders and a numerator of patients with one or more scheduled multimodal analgesia agent.

Endpoints

The primary endpoint of this analysis was the average oral MME used on days 1 through 5 of hospitalization. Days were measured as hours from admission in 24-hour increments. Subgroup analyses of the primary outcome were also completed for surgical and non-surgical groups. Secondary endpoints included percentage of patients with an opioid ordered for analgesia who received a scheduled non-opioid analgesic, percentage of patients with an in-hospital administration of naloxone, length of stay, in hospital mortality, and percentage of patients with nausea, vomiting, constipation, confusion, or bradypnea, defined as less than 12 breaths per minute, reported for each hospital day 1 through 5.

Statistics

Sample size calculations were not performed due to the volume of population health data that was anticipated. Nominal data was analyzed with the Chi-squared test and continuous data was analyzed using the Student’s t-test. Statistical significance was considered at an alpha of less than .05 and all analyses were conducted through JMP^® Version Pro 15 (SAS Institute Inc., Cary, NC, 1989-2019).

Results

The pre- and post-groups included 86 535 patients and 85 194 patients, respectively. Patients in the 2 groups were similar at baseline regarding gender, age, race, presence of an ICD-10 code for opioid toxicity, chronic pain, COPD, and presence of a consult for palliative care (Table 1). The post-group had more patients with an ICD-10 code for opioid use or dependence (4.4% vs 3.7%; P < .0001), diabetes mellitus (27.3% vs 26.0%; P < .0001), heart failure (16.0% vs 14.1%; P < .0001), consult with a pain management specialist (4.8% vs 1.8%; P < .0001), and 1 or more surgical procedures on days 1 through 5 of admission (30% vs 28.5%; P < .0001). The most commonly reported pain locations varied between groups with a larger portion of the pre-group visits reporting abdominal pain and more patient visits in the post-group reporting back and joint pain (Table 1). When stratified for hospital length of stay, there was a reduction in average oral MME used on days 1 through 5 between groups (P < .0001) (Table 2). In the subgroup analysis of patient visits without a surgical procedure in the first 5 days, the reduction in average oral MMEs in the post-group also achieved statistical significance (P < .0001) (Table 3). The subgroup of surgical patients also showed a statistically significant reduction in average oral MMEs in the post-group (P < .0001) (Table 4). The post-group had less documented nausea on hospital days 1 through 5 per patient visit compared to the pre-group (P < .008) (Table 5). Additionally, vomiting and constipation were lower on hospital day 1 in the post-group. Utilization of at least one scheduled multimodal analgesia agent increased from 33% in January of 2016 to 49% in December of 2019 (Figure 2). Average length of stay was 4.2 ± 5.2 days in the post implementation group compared to 4.3 ± 5.7 days in the pre implementation group (P = .0018). Lastly, the outcomes of in hospital naloxone use (P = .08) and mortality (P = .71) were not found to be statistically significant between groups.

Table 1.

Demographics and Clinical Characteristics at Baseline.

Characteristic	Pre (n = 86 535)	Post (n = 85 194)	P-value
Age (mean ±SD)	61 ± 19.2	61 ± 18.9
Female (no. (%))	49 605 (57.3)	49 171 (57.7)
Race (no. (%))
American Indian/Alaska	346 (0.4)	292 (0.3)
Asian/Pacific Islander	638 (0.7)	558 (0.7)
Black/African American	7617 (8.8)	7349 (8.6)
Caucasian	71 334 (82.4)	66 027 (77.5)
Hispanic/Black	459 (0.5)	470 (0.6)
Hispanic/Caucasian	5171 (6.0)	4933 (5.8)
Not reported	395 (0.5)	5065 (6.0)
Other	575 (0.7)	500 (0.6)
Chronic opioid use or dependence (%)	3227 (3.7)	3716 (4.4)	<.00001
Opioid toxicity (%)	1003 (1.2)	981 (1.2)	.89
Chronic pain (%)	12 653 (14.6)	12 617 (14.8)	.27
Past medical history (%)
Diabetes mellitus	22 502 (26.0)	23 291 (27.3)	<.0001
Heart failure	12 213 (14.1)	13 652 (16.0)	<.0001
COPD	15 020 (17.4)	15 056 (17.7)	.09
Consult for pain management (%)	1553 (1.8)	4086 (4.8)	<.0001
Consult for palliative care (%)	2970 (3.4)	2953 (3.5)	.66
Surgical procedure during the first 5 day of hospitalization (%)	24 655 (28.5)	25 527 (30.0)	<.0001
Pain location (%)
Chest	8090 (9.35)	7862 (9.2)	.39
Back	5993 (6.93)	6466 (7.59)	<.0001
Abdominal	4136 (4.78)	3847 (4.52)	.01
Joint	1753 (2.03)	1891 (2.22)	.0054
Limb	1031 (1.19)	998 (1.17)	.704

Table 2.

Entire Hospital Adult Population Analysis of Oral MME Used on Days 1 Through 5.

	Pre	Post	P-value
Day 1	(n = 86 535)	(n = 85 194)
MME	32.6 ± 55.4	27.9 ± 49.7	<.0001
Day 2	(n = 71 718)	(n = 69 363)
MME	34.3 ± 61.6	29.7 ± 57.6	<.0001
Day 3	(n = 48 404)	(n = 47 003)
MME	36.5 ± 65.7	32.4 ± 60.8	<.0001
Day 4	(n = 33 964)	(n = 32 829)
MME	37.5 ± 67.9	33.7 ± 62.9	<.0001
Day 5	(n = 25 114)	(n = 24 178)
MME	38.2 ± 68.7	34.8 ± 65.7	<.0001

Table 3.

Non-surgical Subgroup Analysis of Oral MME Used on Days 1 Through 5.

	Pre	Post	P-value
Day 1	(n = 61 880)	(n = 59 667)
MME	29.7 ± 55.9	25.8 ± 51.5	<.0001
Day 2	(n = 55 577)	(n = 53 538)
MME	30.5 ± 61.1	27.1 ± 58.6	<.0001
Day 3	(n = 38 937)	(n = 37 268)
MME	33.7 ± 65.7	29.7 ± 61.4	<.0001
Day 4	(n = 26 587)	(n = 25 128)
MME	34.8 ± 67.4	31.1 ± 63.9	<.0001
Day 5	(n = 19 159)	(n = 17 718)
MME	35.5 ± 68.3	32.4 ± 67.4	<.0001

Table 4.

Surgical Subgroup Analysis of Oral MME Used on Days 1 Through 5.

	Pre	Post	P-value
Day 1	(n = 24 655)	(n = 25 527)
MME	39.9 ± 53.4	32.9 ± 45.0	<.0001
Day 2	(n = 22 285)	(n = 22 657)
MME	43.6 ± 61.8	35.7 ± 54.7	<.0001
Day 3	(n = 16 663)	(n = 16 208)
MME	43.0 ± 65.1	38.5 ± 58.8	<.0001
Day 4	(n = 11 815)	(n = 11 958)
MME	43.4 ± 68.7	39.1 ± 60.5	<.0001
Day 5	(n = 8698)	(n = 9058)
MME	44.1 ± 69.4	39.5 ± 62.0	<.0001

Table 5.

Frequency of ORADEs on Hospital Days 1 Through 5.

	Pre (no. (%))	Post (no. (%))	P-value
ORADEs Day 1	n = 86 535	n = 85 194
Nausea	7868 (9.1)	6185 (7.3)	<.0001
Vomiting	2798 (3.2)	2121 (2.5)	<.0001
Constipation	2986 (3.5)	2427 (2.9)	<.0001
Confusion	6937 (8.0)	6800 (8.0)	.8
Bradypnea	50 (0.1)	54 (0.1)	.7
ORADEs Day 2	n = 71 718	n = 69 363
Nausea	4716 (6.6)	3878 (5.6)	<.0001
Vomiting	1201 (1.7)	1140 (1.6)	.7
Constipation	3438 (4.8)	3138 (4.5)	.02
Confusion	6725 (9.4)	6720 (9.7)	.05
Bradypnea	40 (0.1)	43 (0.1)	.7
ORADEs Day 3	n = 48 404	n = 47 003
Nausea	2937 (6.1)	2457 (5.2)	<.0001
Vomiting	746 (1.5)	657 (1.4)	.09
Constipation	3963 (8.2)	3800 (8.1)	.7
Confusion	6008 (12.4)	6042 (12.9)	.05
Bradypnea	24 (0.0)	33 (0.1)	.2
ORADEs Day 4	n = 33 964	n = 32 829
Nausea	2017 (5.9)	1589 (4.8)	<.0001
Vomiting	510 (1.5)	433 (1.3)	.07
Constipation	3408 (10.0)	3381 (10.3)	.2
Confusion	4944 (14.6)	4933 (15.0)	.1
Bradypnea	22 (0.1)	17 (0.1)	.5
ORADEs Day 5	n = 25 114	n = 24 178
Nausea	1348 (5.4)	1119 (4.6)	.008
Vomiting	372 (1.5)	309 (1.3)	.08
Constipation	2553 (10.2)	2594 (10.7)	.02
Confusion	3917 (15.6)	3877 (16.0)	.3
Bradypnea	13 (0.1)	24 (0.1)	.07

Figure 2.

Percentage of patients with an opioid ordered and one or more scheduled multimodal analgesia agents over time.

Discussion

Despite the importance of opioid stewardship, hospitals may encounter resource limitations during the design and implementation of effective stewardship programs as described previously by Santalo. ⁹ Development and implementation of a multimodal analgesia order set has the potential to be a resource efficient component of an opioid stewardship program while promoting best practices from the Society of Hospital Medicine and the 2019 Clinical Guidelines for Pain Management in Acute Musculoskeletal Injury.^4,5 The benefits of scheduled multimodal analgesia have been previously explored in surgical and trauma populations; however, our study demonstrates these benefits in a whole hospital system adult population. In the perioperative phase of total hip/knee arthroplasties, Memtsoudis and colleagues found that as the number of non-opioid analgesic medications increased the total opioid usage and ORADEs decreased. ¹⁰ Use of oral multimodal analgesia also resulted in an oral MME per patient day decrease of 31% in the trauma population and led to lower patient reported pain scores. ¹¹ The aforementioned studies support the use of multimodal analgesia as a method to reduce potential harm from opioid use while maintaining patient comfort in surgical and trauma patient populations. This research bridges the important gap in literature surrounding the effects of multimodal analgesia on hospitalized non-surgical adult populations.

Our research has limitations due to its retrospective nature, including reliance on accurate documentation in the medical record. This is especially relevant for the outcome of ORADEs as the accuracy of this outcome depends on consistent documentation of subjective symptoms by nursing staff over the 4-year study period. Additionally, ORADEs reported in this research data set may be due to non-opioid related disease processes and are subject to significant confounding. Due to limitations of the custom reports, patients managed on multimodal analgesia alone were not captured in this analysis which has the potential to reduce the effect shown. The presence of a surgical procedure was defined as a procedure conducted in the operating room. Surgical patient visits were not stratified by the procedure performed; however, the 10 most commonly performed procedures for the pre and post groups can be found in Supplemental Table 1. Bedside procedures or those completed in the emergency department were excluded although these procedures may contribute to significant pain. Additionally, pain severity and frequency of each multimodal analgesia regimen were unavailable for collection and analysis. Patients discharged prior to hospital day 5 were removed from subsequent days of data collection potentially creating selection bias for patients with a higher severity of illness and higher use of oral MME as hospital length of stay increased. In the state of Florida effective July 2018, legislation required continuing education courses for outpatient controlled substance prescribing and placed quantity limits on outpatient opioid prescribing for acute pain. This may have confounded the study results due to impacting prescriber habits and perceptions. Lastly, due to the large sample size included in this analysis, the study is at risk of being overpowered to find a statistically significant difference when no clinically meaningful difference exists.

Future research in the area of multimodal pain management in hospitalized adult populations is needed to better define the most effective multimodal analgesia regimen for implementation on an individual patient level. The development of opioid order sets which promote the use of multimodal analgesia prior to opioid ordering may increase provider awareness of multimodal analgesia options, although this may not directly correlate to an increase in use. Pharmacists are in a unique position to become opioid stewards through promotion and recommendation of multimodal analgesia to prescribers. Pharmacy surveillance systems can be designed to notify pharmacists and prescribers of patients receiving high daily oral MMEs who are candidates for multimodal analgesia. This creates potential for the adoption of hospital protocols for pharmacists to independently order scheduled multimodal analgesia agents.

Conclusion

This retrospective analysis showed the association of a reduction in oral MME after implementation of a multimodal analgesia order set in a multihospital health-system. These results remained consistent for surgical and non-surgical populations thereby suggesting the benefit of multimodal analgesia in reducing daily oral MME in whole hospital adult populations. This analysis supports further use of multimodal analgesia as a reasonable option for inclusion in opioid stewardship initiatives to reduce opioid utilization in a broad population of hospitalized patients.