Economic Impact of Preoperative Meloxicam IV Administration in Total Knee Arthroplasty: A Randomized Trial Sub-Study

Abstract

We evaluated the economic impact associated with preoperative meloxicam IV 30 mg vs placebo administration among adult total knee arthroplasty (TKA) recipients enrolled in Phase IIIB NCT03434275 trial. Data on total hospital costs and length of stay (LOS) obtained from the trial were compared between meloxicam IV 30 mg and placebo groups. Patients in the meloxicam IV 30 mg vs placebo group (n = 93 vs 88) incurred an adjusted $2,266 (95% CI: -$1,035, $5,116; p = 0.1689) lower total hospital costs and an adjusted 8.6% (95% confidence interval [CI]: −2.0%, 18.1%; p = 0.1082) shorter LOS. While statistically non-significant, based on 95% CIs, the results from this sub-study may suggest a favorable impact associated with meloxicam IV 30 mg on hospital costs and LOS.

KEYWORDS:

• Total knee arthroplasty
• meloxicam IV
• total hospital costs
• length of stay
• multimodal pain management

Summary points

• Clinical studies of meloxicam IV have previously evaluated post-surgical dosing to quantify reductions in pain severity. Research on preoperative administration of meloxicam IV 30 mg and associated clinical and economic outcomes among patients undergoing total knee arthroplasty (TKA) has not been previously conducted.

• We evaluated the economic impact associated with preoperative meloxicam IV 30 mg vs placebo administration as part of a modified multimodal pain management plan and associated clinical endpoints among adult TKA recipients enrolled in a Phase IIIB clinical trial.

• Data were obtained from a multicenter Phase IIIB randomized double blind placebo-controlled trial in which adult subjects aged 35-80 years (inclusive) scheduled to undergo an elective primary unilateral TKA were randomized 1:1 to meloxicam IV 30 mg or placebo, as part of a modified multimodal pain management plan.

• Total hospital costs, length of stay (LOS) and 30-day post discharge healthcare resource utilization (HRU) measures obtained from the clinical trial were described and compared for meloxicam IV 30 mg and placebo groups.

• Impact associated with meloxicam IV 30 mg (vs placebo) administration and the clinical endpoints on total hospital costs and LOS were analyzed using generalized linear models (GLM), adjusted for demographic and surgical variables.

• Total mean hospital costs and LOS were numerically lower in the meloxicam IV 30 mg (n = 93) vs. placebo (n = 88) group ($29,302 vs. $31,821; p = 0.7037 and 2.05 vs. 2.24 days; p = 0.3887, respectively). A lower proportion of patients in the meloxicam IV 30 mg group had 30-day hospital readmissions, emergency visits and discharge to skilled nursing facility as compared to the placebo group (1.1% vs. 3.4%; 0.0% vs. 4.5%; 5.4% vs. 14.8%, respectively).

• Adjusted analyses showed that patients in the meloxicam IV 30 mg group incurred $2,266 (95% CI: -$1,035, $5,116) lower total hospital costs (p = 0.1689) and had 8.6% (95% CI: -2.0%, 18.1%) shorter LOS vs. the placebo group (p = 0.1082).

• Every unit (1 mg) increase in opioid consumption was associated with a 0.5% (95% CI: 0.3%, 0.8%) increase in LOS (p = 0.0001). Patients with vs. without ≥1 opioid-related adverse drug events (ORADEs) had 15.1% (95% CI: 2.4%, 29.3%) higher LOS (p = 0.0001). One unit increase in patient global assessment of pain score was associated with a 7.2% (95% CI: 2.7%, 11.6%) decrease in LOS (p = 0.0024).

• The results from this economic sub-study of Phase IIIB placebo-controlled trial indicated that preoperative meloxicam IV 30 mg use may be associated with decreased total hospital costs, shorter LOS, lower 30-day HRU in addition to reduced opioid use and related AEs.

• The sub-study was not sufficiently powered to detect significant differences for economic endpoints between meloxicam IV 30 mg and placebo groups. Preliminary results from our study suggest a need for future studies that are adequately powered to further evaluate the impact associated with meloxicam IV 30 mg on short-term and long-term HRU.

Introduction

Total knee arthroplasty (TKA) is one of the most commonly performed orthopedic procedures in the United States (US) (1). More than half of TKA recipients report extreme pain immediately following surgery and less than half report adequate pain relief with post-operative pain management (2, 3). Inadequate pain relief not only leads to increased suffering but can also result in a myriad of issues, including but not limited to, higher risk of cardiovascular complications, decreased pulmonary function and thrombus formation, increased risk of infections and chronic pain (4–6). Inadequate pain management can therefore negatively affect post-operative recovery and result in longer hospitalization and higher medical costs (7). Opioids have been the mainstay for post-operative pain management, but they are associated with a risk of addiction, overdose and unwanted side effects such as vomiting, constipation and respiratory depression (8). Pain management guidelines support the use of opioid-sparing strategies for post-operative pain management (9). Multimodal approaches that use two or more analgesic agents with different mechanisms of action are recommended to enhance analgesia, minimize opioid consumption and avoid opioid-related adverse events (ORADEs) (9–11). A study conducted using all-payer, US hospital-based database showed variations of 8% to 92% in the use of nonopioid multimodal analgesic therapy after adjusting for demographic and clinical characteristics (12).

Perioperative use of nonsteroidal anti-inflammatory drugs (NSAIDs) has been shown to be safe and effective in reducing perioperative pain, decreasing opioid requirements, and in minimizing ORADEs (13). While there is little recent peer-reviewed evidence, several studies conclude that the use of multimodal analgesia, including NSAIDs, does result in a shorter length of stay (LOS) which could subsequently reduce total costs (14–16). In addition, minimal research has been conducted specific to adverse events (AEs) or serious AEs (SAEs) with other NSAIDs in multimodal analgesia studies in TKA. In the research that has been published for non-knee specific arthroplasties, both peri- and postoperative, there are advantages with the main goal of reducing ORADEs (17, 18). A study conducted in China showed that early perioperative use of meloxicam had better pain control compared with postoperative administration among patients that had an arthroscopic knee surgery (19). For perioperative use, between oral and intravenous (IV) meloxicam, IV meloxicam is preferred for the treatment of acute pain as it can achieve faster onset of action and peak plasma concentrations compared to oral formulations (20–22).

Intravenous meloxicam (ANJESO^TM) is a novel NSAID formulation that utilizes nanocrystal technology and is indicated for use in adults for the management of moderate-to-severe pain, alone or in combination with non-NSAID analgesics. Several prior phase II/III clinical trials have demonstrated favorable safety and efficacy profiles and reductions in pain with post-operative meloxicam IV administration (23–25); however, data on preoperative meloxicam IV administration and associated economic outcomes were not collected. Patients receiving multimodal analgesia were associated with decrease in length of stay of up to −9.3% in knee arthroplasty for those who received more than 2 modes of analgesics in addition to opioids. However, there were not similar levels in reduction of the hospitalization costs (−1.3% for knee arthroplasty for those who received more than 2 modes of analgesics in addition to opioids) (26).

A Phase IIIB multicenter randomized placebo-controlled trial (https://clinicaltrials.gov/ct2/show/NCT03434275) was conducted to assess the efficacy and safety of preoperative administration of meloxicam IV 30 mg as part of a modified multimodal pain management plan among patients undergoing TKA. As part of this study, inpatient healthcare resource utilization (HRU) and cost data were also captured. The objectives of this sub-study were to evaluate the treatment impact associated with preoperative meloxicam IV 30 mg compared to placebo on total costs of hospital stay and LOS among patients undergoing TKA and to evaluate impact associated with clinical end points including opioid consumption, safety, and efficacy on total costs of hospital stay and LOS. Additionally, hospital readmissions, emergency room (ER) visits and requirement of skilled nursing facility (SNF) care within 30 days post hospital discharge were compared between the meloxicam IV 30 mg and placebo groups.

Materials and methods

Study design and participants

A prospective economic sub-study was performed alongside a multicenter Phase IIIB randomized double blind placebo-controlled trial in which adult subjects aged 35-80 years (inclusive) scheduled to undergo an elective primary (no repeat arthroplasties) open unilateral TKA between February 19, 2018 to May 21, 2019 were randomized 1:1 to meloxicam IV 30 mg or placebo. Screening occurred within 28 days before undergoing surgery on Day 1. (Additional details on inclusion and exclusion available at https://clinicaltrials.gov/ct2/show/NCT03434275). As part of multimodal pain management, all subjects received oral acetaminophen 650 mg and oral gabapentin 600 mg 30 to 90 minutes prior to the surgery, in addition to appropriate prophylactic IV antibiotic and tranexamic acid IV 1 gm. All subjects continued to receive acetaminophen 650 mg through last study dose plus 1 day (27). Doses of the study drug were administered by staff members blinded to treatment assignment. The first dose of blinded-study drug was administered as an IV bolus after the start of spinal anesthesia and prior to surgery, then repeated every 24 hours thereafter until no longer required or IV access was removed. Subjects were able to utilize opioids postoperatively for analgesia as required, though no additional NSAIDs were allowed. Data on safety, tolerability (opioid related adverse drug events (ORADEs)), and efficacy were obtained in addition to HRU (total hospital costs, hospital LOS, hospital readmissions, ER visits, doctor visits, requirement of SNF care and doctors office calls due to pain). Staff members performing subject assessments were blinded to the treatment. The study was conducted under an FDA Investigational New Drug and IRB (Western Institutional Review Board; WIRB) approval was obtained prior to study conduct. Written informed consent was obtained from all study participants. A subject-level database was developed to capture quantity of service, charges and date of service from UB-04 claims forms (from admission date to discharge).

Study outcomes/endpoints

Clinical endpoints

To assess the impact associated with clinical endpoints on total costs of hospital stay and LOS, data on opioid consumption, AEs of special interest defined as selected events related to concerns associated with NSAIDs (e.g. bleeding, cardiovascular, hepatic, injection site reactions, renal, thrombotic, and wound healing event), SAEs (e.g., hepatobiliary, renal, cardiovascular, thrombotic, bleeding, wound healing, injection site events), ORADES (e.g., nausea, vomiting, constipation, stomach pain, loss of appetite, ileus, sleepiness, tiredness, drowsiness, dizziness, lightheadedness, weakness, itching, dry mouth, respiratory depression, apnea and respiratory arrest) and efficacy were captured from the clinical trial data. Opioid consumption was collected in the clinical trial data and converted to the intravenous morphine equivalent dose (IVMED, mg) and calculated as total use of opioid analgesia from hour 0 through hospital discharge. Efficacy was assessed as sum of pain intensity and as pain control. Sum of pain intensity was reported as the time of first dose of study drug through 24 hours as well as the sum of pain intensity through 48 hours and from first dose of the study drug through discharge. Pain control assessed on post-operative day (POD) 1, 2 and through hospital discharge was measured using the patient global assessment (PGA) scale, a 5-point scale ranging from poor (0) to excellent (5). Clinical endpoints including overall benefit of analgesia score (OBAS) derived from 7 items ranging from not at all (0) to very much (4, 28) and opioid dimension distress scale (ODDS) subdimension scores measured using a 4-point scale with 3 symptom distress dimensions (frequency, severity, bothersomeness) for 12 symptoms (29), were also assessed and reported on POD 1.

Total costs of hospital stay and LOS

Total costs of hospital stay included all charges incurred from admission date until discharge date. These included private room, drug costs, medical/surgical supplies, sterile supply, laboratory, pathology lab, operative room, anesthesia, recovery room, physical therapy, respiratory therapy, diagnostic, occupational therapy and pulmonary function costs. A national cost:charge ratio of 0.477 from the Healthcare Cost and Utilization Project (HCUP) was applied to convert ‘charges’ to ‘costs’(30). Total costs were reported for meloxicam IV 30 mg and placebo groups and stratified by sparing of quadriceps tendon during surgery, insurance type, presence of AEs of special interest, presence of ORADEs and by presence of SAEs.

Hospital LOS was calculated as number of days and number of hours from date of admission to the date of actual hospital discharge and reported for meloxicam IV 30 mg and placebo groups and stratified by presence of AEs of special interest, presence of ORADEs and by presence of SAEs.

30-Day Post discharge healthcare resource utilization

Thirty-day post discharge HRU was defined as hospital readmissions, ER visits, physician office visits, phone calls due to pain and requirement of SNF care after discharge, which were measured from the day of hospital discharge through POD 30.

Study size

The planned sample size for the clinical trial (100 patients per group) had at least 90% power to detect the difference between meloxicam IV 30 mg vs placebo group in total opioid consumption based on results from a Phase III safety study that evaluated all major surgeries (25). As statistical power calculated before data collection was based on information on opioid consumption obtained from previous studies to determine the sample size needed for the current study, it may not take into account the adequate power needed to detect between group differences for economic endpoints. Therefore, for this sub-study, a post-hoc power analysis was conducted based on the results of the study, to determine whether any non-significant results were due to lack of statistical power (31, 32).

Statistical analysis

Patient demographics were reported as mean, median and standard deviation (SD) for continuous variables and as frequency and percentage for categorical variables. Total costs of hospital stay, hospital LOS, opioid consumption, sum of pain intensity, PGA score, OBAS and ODDS scores were summarized as mean, median and SD for the meloxicam IV 30 mg and placebo groups. Presence and number of AEs were summarized as frequencies and percentages for the meloxicam IV 30 mg and placebo groups. After assessing the data distribution, as data for clinical endpoints including, sum of pain intensity, PGA scores, OBAS, and ODDS scores were normally distributed, p-values were derived using independent sample t-tests. Data for opioid consumption and economic endpoints (total costs of hospital stay and LOS) were not normally distributed; therefore, Wilcoxon rank-sum tests were used to assess differences in these continuous variables while chi-square tests were used to assess the difference in the proportion for categorical variables between the two groups. Fisher’s exact test was used where cell sizes were less than 5.

To assess the treatment impact as well as to assess the impact associated with opioid consumption, presence of ORADEs, pain control, OBAS and ODDS scores on total hospital costs and hospital LOS (in days), generalized linear models (GLMs) with log link and gamma distribution were used, adjusting for other covariates of interest. Other covariates of interest included age group (≥65 vs. <65 years), gender (male vs. female), ethnicity (Hispanic or Latino vs. not Hispanic or Latino) and whether quadriceps tendon spared during surgery (yes vs. no). Multicollinearity was evaluated during model development. We did not include body mass index (BMI) as an independent variable in the above GLM models, except when assessing the impact of pain control on economic outcomes as BMI was affecting model fit due to high collinearity with other independent variables of interest. GLM results are expressed as adjusted exponentiated parameter estimates, their 95% confidence intervals (CIs) and associated p-values.

Results on hospital readmissions, ER and doctor visits, phone calls due to pain and requirement of SNF care obtained during the clinical trial were summarized and reported for the meloxicam IV 30 mg and placebo groups. P-values were not computed for these endpoints due to very small sample sizes for proportion of patients with these HRU.

All analyses were conducted using SAS^® version 9.4 (SAS Institute Inc., Cary, NC).

Results

A total of 181 subjects were enrolled in the trial, 93 in the meloxicam IV 30 mg group and 88 in the placebo group. The groups were comparable in terms of demographic characteristics (60% aged ≥65 years, 58% females, and nearly 80% white in both treatment groups). In both groups, majority of the subjects had private insurance (54.8% and 52.3%, respectively). Four subjects (2.2%) had ‘other’ insurance which included CHAMPVA, Medicaid, Medicare/United Health and Medicare/Medicaid (1 patient each) (Table 1).

Table 1. Enrolled subjects.

Measures of interest	Meloxicam IV		Placebo		p-value^b
	N	%	N	%
All Enrolled Subjects	93	51.4%	88	48.6%
Age (years)
Mean (SD)	66.9 (8.23)		65.5 (8.11)		0.2730
Age ≥65 years	58	62.4%	54	61.4%	0.8897
Sex					0.9880
Male	39	41.9%	37	42.0%
Female	54	58.1%	51	58.0%
Race					0.9992
White	74	79.6%	70	79.5%
Black/African American	18	19.4%	17	19.3%
Asian	1	1.1%	1	1.1%
Ethnicity					0.9017
Hispanic or Latino	9	9.7%	9	10.2%
Not Hispanic or Latino	84	90.3%	79	89.8%
BMI (kg/m^2 )
Mean (SD)	30.6 (4.74)		31.5 (5.05)		0.1991
Insurance type					0.7294
Private	51	54.8%	46	52.3%
Medicare/Other^a	42	45.2%	42	47.7%
Was the quadriceps tendon spared during surgery?					0.3722
Yes	67	72.0%	58	65.9%
No	26	28.0%	30	34.1%

BMI, Body mass index; SD, Standard deviation.

^a Only 4 patients had "other" health insurance which includes CHAMPVA (n = 1), Medicaid (n = 1), Medicare/United health (n = 1) and Medicare/Medicaid (n = 1).

^b p-values for the comparison between meloxicam IV and placebo groups were obtained using chi-square tests.

Total mean costs of hospital stay were lower by 7.9% in the meloxicam IV 30 mg group than the placebo group; however, the difference was not statistically significant ($29,302 vs. $31,821; p = 0.7037). The largest driver of the mean difference in total hospital costs between the meloxicam IV 30 mg and placebo group were operative room costs ($8,038 ±$4,722 vs. $9,295 ±$11,300), followed by respiratory therapy costs ($207 ±$292 vs. $675 ±$885) and diagnostic costs ($301 ±$132 vs. $818 ±$3,283) (findings not presented in tabular form). Following stratification, total mean costs were higher in patients with Medicare/other insurance compared to patients with private insurance in the meloxicam IV 30 mg group ($32,011 vs. $27,072), whereas in placebo group, patients with Medicare/other insurance and patients with private insurance incurred similar mean total costs of hospital stay ($31,459 vs. $32,152). Total mean costs were numerically higher among patients whose quadriceps tendon was spared during surgery compared to those whose quadriceps tendon was not spared during surgery in both meloxicam IV 30 mg ($30,659 vs. $25,806) and placebo groups ($34,842 vs. $25,981). Upon stratification, none of the cost comparisons between the two treatment groups were significantly different (Table 2).

Table 2. Total Costs of Hospital Stay.

Measures of interest	Meloxicam IV (N = 93)			Placebo (N = 88)
	mean	SD	median	mean	SD		median	p-value^f
Total costs of hospital stay^a	$29,302	$15,914	$24,451	$31,821	$22,281		$25,273	0.7037
Total costs of hospital stay by^a
Insurance type
Private	$27,072	$11,671	$24,834	$32,152	$21,237		$30,092	0.3608
Medicare/Other^b	$32,011	$19,709	$24,249	$31,459	$23,624		$23,456	0.6007
Whether quadriceps tendon was spared during surgery
Yes	$30,659	$15,220	$30,714	$34,842	$21,225		$31,306	0.3469
No	$25,806	$17,399	$20,337	$25,981	$23,458		$19,892	0.582
AEs of special interest^c
0	$29,633	$16,328	$24,451	$31,516	$21,304		$25,273	0.8236
≥1	$24,501	$6,689	$23,591	$33,756	$28,759		$25,011	0.7431
ORADEs^d
0	$29,612	$17,041	$24,266	$29,916	$17,833		$24,129	0.991
≥1	$28,971	$14,803	$26,175	$32,620	$23,989		$25,980	0.7262
SAEs^e
0	$29,307	$16,074	$24,368	$29,604	$19,188		$24,655	0.9498
≥1	$29,151	$12,331	$35,978	$51,288	$36,526		$34,877	0.5791

AEs, Adverse events; ORADEs, Opioid related adverse drug events; SAEs, Serious adverse events; SD, Standard deviation.

^a A cost to charge ratio (0.477) was applied to convert the charges obtained from UB-04 forms to cost.

Note: Readmission was not captured in the UB-04 form; therefore, charges from UB-04 do not reflect readmission.

^b Only 4 patients have "other" health insurance which includes CHAMPVA (n = 1), Medicaid (n = 1), Medicare/United health (n = 1) and Medicare/Medicaid (n = 1).

^c AEs of special interest refer to the adverse events related to the study drug only (e.g. bleeding, cardiovascular, hepatic, injection site reactions, renal, thrombotic, and wound healing event).

^d ORADEs include adverse events such as gastrointestinal (e.g. nausea, vomiting, constipation, stomach pain, loss of appetite, ileus), central nervous system (e.g. sleepiness, tiredness, drowsiness, dizziness, lightheadedness, weakness, itching, dry mouth) or respiratory effects (e.g. respiratory depression, apnea, respiratory arrest).

^e SAEs include hepatobiliary, renal, cardiovascular, thrombotic, bleeding, wound healing or injection site events.

^f p-values used for comparison between treatment groups were derived using Wilcoxon rank-sum test. P-values for variables with small cell sizes should be interpreted with caution.

Mean hospital LOS in days was numerically lower among subjects in meloxicam IV 30 mg than placebo groups; however, the differences were not statistically significant (2.05 vs. 2.24 days; p = 0.3887). It was observed that among patients with at least one SAE, hospital LOS in days was lower among meloxicam IV 30 mg group than placebo group (2.67 vs. 3.44; p = 1.000). The same was observed for hospital LOS in hours (71.70 vs. 92.34; p = 0.7115), although the comparisons were not significant (Table 3).

Table 3. Hospital Length of Stay (LOS).

Measures of interest	Meloxicam IV (N = 93)			Placebo (N = 88)
	mean	SD	median	mean	SD	median	p-value^e
Hospital LOS in days^a	2.05	0.7	2	2.24	1.18	2	0.3887
Hospital LOS in hours^a	56.35	16.98	55.5	61.19	29.02	55.6	0.4502
Number of days from the date of admission to the date when the hospital discharge order was written	1.97	0.7	2	2.14	1.18	2	0.5933
Hospital LOS (in days) by
AEs of special interest^b
0	2.05	0.71	2	2.25	1.23	2	0.3674
≥1	2.17	0.41	2	2.17	0.83	2	0.9108
ORADEs^c
0	1.96	0.74	2	1.81	0.69	2	0.4526
≥1	2.16	0.64	2	2.42	1.3	2	0.3594
SAEs^d
0	2.03	0.68	2	2.1	0.74	2	0.519
≥1	2.67	1.15	2	3.44	2.83	2	1
Hospital LOS (in hours) by
AEs of special interest
0	56.17	17.37	55.5	61.5	30.23	56.2	0.3944
≥1	58.93	10.52	55.2	59.17	20.67	54	0.6734
ORADEs
0	54.04	17.71	55.4	50.76	16.76	53.7	0.4481
≥1	58.82	16	55.8	65.56	31.94	56.8	0.3995
SAEs
0	55.84	16.66	55.1	57.64	17.85	55.2	0.5781
≥1	71.7	23.5	58.4	92.34	69.53	58	0.7115

AEs, Adverse events; LOS, Length of stay; ORADEs, Opioid-related adverse events; SAEs, serious adverse events; SD, Standard deviation.

^a LOS (in hours and days) was reported as number of hours from hospital admission to discharge (captured in clinical database) as well as the number of days from the date of admission to the date of actual hospital discharge.

^b AEs of special interest refer to the adverse events related to the study drug only (e.g. bleeding, cardiovascular, hepatic, injection site reactions, renal, thrombotic, and wound healing event).

^c ORADEs include adverse events such as gastrointestinal (nausea, vomiting, constipation, stomach pain, loss of appetite, ileus), central nervous system (sleepiness, tiredness, drowsiness, dizziness, lightheadedness, weakness, itching, dry mouth) or respiratory effects (respiratory depression, apnea, respiratory arrest).

^d SAEs include hepatobiliary, renal, cardiovascular, thrombotic, bleeding, wound healing or injection site events.

^e p-values used for comparison between treatment groups were derived using Wilcoxon rank-sum test. P-values for variables with small cell sizes should be interpreted with caution.

Results on clinical endpoints estimated from the clinical trial data are reported in Table 4. Mean total use of opioid analgesia from hour 0 to discharge was significantly lower among meloxicam IV 30 mg group compared to placebo group (33.28 vs. 44.87 mg; p < 0.001). Proportion of patients with ≥1 ORADEs were significantly higher for placebo group than the meloxicam IV 30 mg group (70.5% vs. 48.4%; p = 0.003). Six patients (6.5%) in meloxicam IV 30 mg group had ≥1 AEs of special interest in comparison to 12 patients (13.6%) in the placebo group. SAEs were observed among 3 patients (3.2%) in meloxicam IV 30 mg group and 9 patients (10.2%) among the placebo group. Mean PGA score from Hour 0 to 24 was significantly higher in the meloxicam IV 30 mg group compared to placebo group (2.52 vs. 1.88; p = 0.0002) and until Hour 48 (2.85 vs. 2.38; p = 0.0025). The mean OBAS score on POD1 was significantly lower in the meloxicam IV 30 mg group compared to placebo group (4.38 vs. 5.87; p = 0.0019). However, mean ODDS score was similar between the two groups (p = 0.06) (Table 4).

Table 4. Clinical endpoint results from Phase IIIB clinical trial by treatment group.

Measures of interest	Meloxicam IV (N = 93)		Placebo (N = 88)		p-value^f
	n/mean (SD)	%/median	n/mean (SD)	%/median
Total use of opioid analgesia from Hour 0 through hospital discharge	33.28 (21.34)	27.7	44.87 (21.76)	44.42	<.0001
Total use of opioid analgesia from hospital discharge through 24-hour telephone interview	5.53 (4.89)	4.6	6.99 (6.12)	5.06	0.1406
AEs of special interest^a
≥1 AEs of special interest	6	6.5%	12	13.6%	0.106
Number of AEs of special interest					0.233
0	87	93.5%	76	86.4%
1	4	4.3%	4	4.5%
2	1	1.1%	5	5.7%
3+	1	1.1%	3	3.4%
ORADEs^b
≥1 ORADES	45	48.4%	62	70.5%	0.003
Number of ORADEs					0.004
0	48	51.6%	26	29.5%
1	23	24.7%	38	43.2%
2	13	14.0%	20	22.7%
3+	9	9.7%	4	4.5%
SAEs^c
≥1 SAEs	3	3.2%	9	10.2%	0.058
Number of SAEs
0	90	96.8%	79	89.8%	0.146
1	3	3.2%	8	9.1%
2+	0	0.0%	1	1.1%
Sum of pain intensity from the time of first dose of study drug through 24 hours	5452.48 (2196.07)	5,299.50	6974.86 (2098.37)	7,184.50	<.0001
Sum of pain intensity from the time of first dose of study drug through 48 hours	10275.11(4321.41)	9,947.50	12720.69 (4267.77)	13,023.00	0.0002
Sum of pain intensity from the time of first dose of study drug through discharge	11989.45 (6901.52)	12,112.00	17141.06(11844.34)	15,220.50	0.0007
PGA score^d (by time)
Hour 0 to 24	2.52 (1.1)	3.0	1.88 (1.07)	2.0	0.0002
Hour 48	2.85 (0.96)	3.0	2.38 (0.84)	2.0	0.0025
Hour 72	2.83 (1.19)	3.0	2.62 (0.74)	3.0	0.5811
OBAS^e score (POD1)	4.38 (2.94)	4.0	5.87 (3.29)	5.0	0.0019
ODDS score (POD1)	1.68 (2.08)	1.0	2.33 (2.54)	2.0	0.0637

AEs, Adverse Events; OBAS, Overall Benefit of Analgesia Score Questionnaire; ODDS, Opioid Dimension Distress Score; ORADEs, Opioid-related adverse events, PGA, Patient Global Assessment; POD, post-operative day; SAEs, Serious adverse events; SD, Standard deviation.

^a AEs of special interest refer to the adverse events related to the study drug only (e.g. bleeding, cardiovascular, hepatic, injection site reactions, renal, thrombotic and wound healing event).

^b ORADEs include adverse events such as gastrointestinal (e.g. nausea, vomiting, constipation, stomach pain, loss of appetite, ileus), central nervous system (e.g. sleepiness, tiredness, drowsiness, dizziness, lightheadedness, weakness, itching, dry mouth) or respiratory effects (e.g. respiratory depression, apnea, respiratory arrest).

^c SAEs include hepatobiliary, renal, cardiovascular, thrombotic, bleeding, wound healing or injection site events.

^d PGA score represents pain control:0-poor, 1-fair, 2-good, 3-very good, or 4-excellent.

^e OBAS score derived from 7 items ranging from 0 = not at all to 4 = very much.

^f As data for sum of pain intensity, PGA scores, OBAS, and ODDS scores was normally distributed, p-values were computed using independent sample t-tests. p-values used for comparison of other continuous variables between treatment groups were derived using Wilcoxon rank-sum test as the data was not normally distributed; p-values for categorical variables were obtained using chi-square tests. P-values were not computed for cell sizes with 0 counts.

Results from GLM showed that after adjusting for age, sex, ethnicity, and whether quadriceps tendon was spared during surgery (yes/no), subjects who were administered meloxicam IV 30 mg incurred 9.8% (95% CI: −4.49%, 22.2%) lower total costs during the hospital stay than subjects in the placebo group, although this difference was not statistically significant (p = 0.1689). Meloxicam IV 30 mg was associated with 8.6% (95% CI: −2.0%, 18.1%) lower LOS as compared to placebo, although the difference was not statistically significant (p = 0.1082). While there was no significant association between opioid consumption and total hospital costs, every unit (1 mg IVMED) increase in opioid consumption from Hour 0 to discharge was associated with a 0.5% (95% CI: 0.3%, 0.8%) significant increase in LOS (p = 0.0001). There was no significant association between presence of ORADEs and total costs of hospital stay. However, subjects with ≥1 ORADEs had 15.1% (95% CI: 2.4%, 29.3%) significantly longer LOS as compared to those with no ORADEs (p = 0.018). There was no significant association between PGA score from 0 to 48 hour and total costs of hospital stay. When assessing association between PGA score from 0 to 48 hour with hospital LOS, we found that one unit increase in PGA score was associated with a significant decrease in LOS (7.2%; 95% CI: 2.7%, 11.6%; p = 0.0024) when other covariates remain constant. Additionally, there was no significant association between OBAS score and total costs of hospital stay or LOS, or ODDS score and total costs of hospital stay or LOS. Due to insufficient sample size and issues related to model fit and convergence, the impact associated with SAEs, AEs of special interest and sum of pain intensity on hospital costs and LOS could not be assessed (Table 5).

Table 5. Multivariable analyses on treatment group and clinical endpoints.

	Total Costs of Hospital Stay				Hospital LOS (days)
	Parameter Estimate	Wald 95% Confidence Limits		p-value	Parameter Estimate	Wald 95% Confidence Limits		p-value
		Lower Limit	Upper Limit			Lower Limit	Upper Limit
Treatment group
Meloxicam IV 30 mg vs Placebo	0.902^c	0.778	1.045	0.1689	0.914	0.819	1.020	0.1082
Clinical endpoints
Opioid consumption (total use of opioid analgesia from Hour 0 through hospital discharge)	0.999	0.996	1.003	0.6480	1.005	1.003	1.008	0.0001
ORADEs^a (≥1 vs 0)	1.039	0.890	1.213	0.6231	1.151	1.024	1.293	0.0183
PGA^b Hour 48	0.996	0.890	1.213	0.9231	0.928	1.024	1.293	0.0024
OBAS score (POD1)	0.984	0.914	1.085	0.2005	1.007	0.884	0.973	0.4924
ODDS (POD1)	0.982	0.960	1.009	0.2940	0.992	0.988	1.026	0.5656

LOS, Length of stay; ORADEs, Opioid-related adverse events; PGA, pain global assessment; OBAS, overall benefit of analgesia; ODDS, opioid distress dimension score.

^a ORADEs include adverse events such as gastrointestinal (nausea, vomiting, constipation, stomach pain, loss of appetite, ileus), central nervous system (sleepiness, tiredness, drowsiness, dizziness, lightheadedness, weakness, itching, dry mouth) or respiratory effects (respiratory depression, apnea, respiratory arrest).

^b PGA score measured from Hour 0 to 48.

^c Note incremental costs of hospital stay were derived using the exponentiated parameter estimate of intercept (exp [10.026] = 23066).

Results on hospital readmissions, ER visits, physician office visits, and phone calls due to pain, and requirement of skilled nursing facility within 30 days post hospital discharge are presented in Table 6. A lower proportion of patients in the meloxicam IV 30 mg vs. placebo group had all-cause hospital readmissions (1.1% vs. 3.4%), ER visits (0.0% vs. 4.5%), phone calls due to pain (4.3% vs. 10.2%) and were discharged to SNF (5.4% vs. 14.8%). None of the patients in either group had unscheduled doctor visits due to pain (Table 6).

Table 6. Healthcare resource utilization measures from hospital discharge through post-operative day 30.

Parameters of Interest	Meloxicam IV (N = 93)			Placebo (N = 88)
	N/mean (SD)	%	Events	N/mean (SD)	%	Events
All-cause hospital readmissions	1	1.1%	1	3	3.4%	3
ER visits due to pain	0	0.0%	0	4	4.5%	4
Unscheduled doctor visits due to pain	0	0.0%	0	0	0.0%	0
Phone calls due to pain	4	4.3%	4	9	10.2%	9
Subjects requiring skilled nursing facility care after discharge	5	5.4%	N/A	13	14.8%	N/A
Total time spent in skilled nursing facility (days)
Mean (SD)	0.6 (3.12)			1.9 (4.99)
Median (Min, Max)	0 (0, 22)			0 (0,22)

ER, Emergency room; SD, Standard deviation.

Discussion

This sub-study demonstrated a favorable impact associated with meloxicam IV 30 mg vs placebo administration, as part of a modified multimodal analgesia, on economic outcomes. Total costs of hospital stay, hospital LOS (in days), proportion of patients with 30-day hospital readmissions, with ER visits and discharged to SNF were numerically lower for the meloxicam IV 30 mg group compared to the placebo group. The between-group differences for mean total hospital costs and LOS did not reach statistical significance, possibly due to lack of power in our study. We observed the study was underpowered with a power of 13.8% and 25.7% to detect a significant difference at a two-sided 95% CI, for mean total hospital costs and LOS, respectively between the two groups. This implies that to achieve at least 80% power for detecting significant differences in mean total hospital costs and LOS between the two groups, we would require a sample size of 958 patients and 410 patients in each group, respectively. Whereas our study had a final sample size of 93 patients in meloxicam IV 30 mg and 88 patients in placebo group, implying that the current study was not sufficiently powered to detect significant differences for economic endpoints. It should be noted that this is a sub-study of a Phase IIIB randomized trial where the planned sample size was calculated to detect clinically meaningful differences in the clinical endpoint of total opioid consumption.

Nevertheless, our study indicated a wide range for the true value of differences in mean total hospital costs (-4.49% to 22.5%) and LOS (-2.0 to 18.1%) between meloxicam IV 30 mg vs placebo group based on 95% CIs. Additionally, a study by Maiese et al. (33) observed a statistically significant difference of $702 in mean total hospital costs (p < 0.0001) when comparing multimodal analgesia including IV acetaminophen with IV opioid monotherapy, among orthopedic surgery recipients. The results of our study consistently indicated a favorable clinical profile for the meloxicam IV 30 mg vs placebo group, administered as part of a modified multimodal analgesia, for all the study outcomes/endpoints, further supporting evidence of a likely favorable impact in terms of total hospital costs and LOS associated with meloxicam IV 30 mg as part of a modified multimodal analgesia. Although, our study did not assess long-term impact associated with perioperative administration of meloxicam IV 30 mg on total costs of care, previous studies have shown a shorter hospital LOS to be associated with lower total costs of care two-year post TKA (34).

The findings related to total hospital costs, LOS and HRU from this sub-study are consistent with the results on clinical endpoints, which showed a significantly lower perioperative use of opioid analgesia and a significantly lower proportion of patients with ≥1 ORADEs in the meloxicam IV 30 mg vs placebo arm (p < 0.0001 and p < 0.001, respectively). Previous studies have shown preoperative opioid use and ORADEs to be associated with longer LOS, higher total hospital costs and increased likelihood of discharge to acute care facilities and of hospital readmissions (35–40). Consistent with previous studies, opioid consumption or presence of ≥1 ORADEs were observed to increase hospital LOS among these patients in our study. While there was no significant association when analyzing the impact associated with opioid consumption or ORADEs on total costs of hospital stay, on average, patients with ≥1 ORADEs incurred $853 more costs than subjects with no ORADE (finding not presented in tabular form).

While our study did not show a significant association between OBAS and ODDS scores on hospital LOS, it should be noted that that due to small sample sizes in this randomized controlled trial, the possibility of residual confounding cannot be ruled out. Previous studies have shown the skillset of surgical team, type of implant and the BMI of patients as predictors of hospital LOS and total hospital costs (40–43). Our study also did not assess the impact associated with surgical complications on hospital LOS and total hospital costs. These limitations of residual confounding also apply to other relationships assessed in our study. Nevertheless, results from this current study showed significantly lower mean PGA, OBAS and ODDS scores in the meloxicam IV 30 mg group, despite the lower mean opioid analgesia use in this group vs the placebo group. These findings, coupled with the results from previous studies (44–46) that show an association between inadequate post-operative pain relief and longer LOS and HRU, are indicative of a potential beneficial impact associated with perioperative meloxicam IV 30 mg administration on total hospital costs and LOS.

Another limitation of this study is that total hospital costs included only those charges that were recorded on the UB-04 forms and billed by the hospital and did not include charges incurred due to professional services. Nevertheless, our study indicated that operative room costs were the largest driver of the mean difference in total hospital costs between the meloxicam IV 30 mg and placebo group, followed by respiratory therapy and diagnostic costs. Further, while this study did not assess post-discharge opioid use, a previous study showed worse pain scores on the day of surgery to be a significant predictor of persistent opioid use over 6 months in patients undergoing TKA or hip arthroplasty (7, 47).

Due to the inclusion/exclusion criteria applied in the Phase IIIB trial, our study sample may not be representative of the general TKA population. Also, due to small sample size, the study lacked sufficient power to detect significant between-group differences. Future research is needed to study the impact of meloxicam IV 30 mg on HRU using a larger sample size.

Conclusions

Safe and effective pain management, while reducing opioid requirements, is one of the primary goals of post-operative care. The results from this economic sub-study of Phase IIIB placebo-controlled trial indicated that preoperative meloxicam IV 30 mg use may be associated with decreased total hospital costs, shorter LOS, lower 30-day HRU in addition to reduced opioid use and related AEs.

Meeting presentations

Material reported in this manuscript was presented at the 45^th Annual Regional Anesthesiology and Acute Pain Medicine Meeting (ASRA); in San Francisco, April 23-25, 2020.

Author contributions

R.D.B., R.S.; and A.P.S., contributed to study design and implementation; managed subjects and helped with data acquisition; helped draft the article and/or critically reviewed it for important intellectual content; approved the final article.

D.S conducted the study analyses. V.R.A., D.S., M.D., K.C., L.K.B, A.F. and W.D. contributed to study design, data analysis plan and reporting of the data/results; provided guidance on data interpretation; helped draft the article and/or critically reviewed it for important intellectual content; approved the final article.

R.J.M., S.W.M., and EC contributed to study design and implementation; provided guidance on data analysis and interpretation; helped draft the article and/or critically review it for important intellectual content; approved the final article.

Acknowledgments

The authors would like to thank Vibha CA. Desai, PhD, Sr. Consultant at IQVIA, for her writing and editorial assistance.

Declaration of interest

V.R. Anupindi, D. Shah, M. DeKoven, K. Coyle – employees of IQVIA and received funding for the study from Baudax Bio, formerly Recro Pharma; S. McCallum, R. Mack, A. Freyer, – Employees and stockholders of Baudax Bio, formerly Recro Pharma. L.K. Black and E. Coyle – Employees of Baudax Bio at the time of study conduct.

Funding

The clinical trial and its publication were supported by Baudax Bio, Inc., Malvern, PA.