Opioid over-prescription after total joint arthroplasty: Findings from patient reported opioid consumption journals

Abstract

We sought to determine postoperative opioid consumption for opioid-naïve patients undergoing total knee and hip arthroplasty (THA and TKA) procedures via an observational cohort study, consisting of 55 patients who underwent either primary unilateral TKA (n = 28) or THA (n = 27). Patients were provided with a journal to track daily consumption of pain medicine. Patients were prescribed an average of 67 opioid pills post-operatively; however, they consumed an average of 31 pills (214 morphine equivalent doses - MED). TKA patients received (p = 0.01) and consumed (p = 0.005) higher amounts of opioids. On average, patients ceased opioid usage at 14 days post-operatively.

1. Introduction

As the current opioid epidemic in the United States continues to escalate, greater emphasis has been placed on prescribers to moderate the use of opioids for pain management following surgical procedures. Orthopedic surgeons are the third highest group of prescribers of opioids, accounting for 11% of all high-volume specialties,¹ yet standardized guidelines regulating opioid prescriptions for orthopedic patients are lacking. For major joint reconstruction surgeries, prescription opioids remain an integral part of acute post-operative pain management; however, prescribing patterns for patients following total knee (TKA) and hip (THA) arthroplasty vary widely, ranging from 10 up to and exceeding 300 opioid pills.2, 3, 4 The lack of standardization often leads to patients receiving prescriptions for a greater quantity of opioid pills than needed4, 5, 6 and risks potential diversion for unused opioids. In order to determine appropriate prescription limits on opioids, details regarding actual usage for patients following joint replacement procedures are needed.

Efforts to both quantify and limit post-operative opioid use in arthroplasty patients have become a crucial area of research.^4,5,7,8 Numerous studies have shown that patients consume substantially fewer opioid pills than they were prescribed, with as high as 80% of pills remaining unused.3, 4, 5, 6 Implementation of restrictive opioid guidelines have resulted in considerable reductions in total morphine equivalents prescribed following joint arthroplasty procedures,^7,8 yet those studies did not assess the effect on consumption rates. In contrast, a recent randomized controlled trial revealed that reducing the quantity of opioids prescribed at discharge resulted in patients consuming less opioids during the first 30 days post-operative⁵; however, requests for refill prescriptions were higher for patients who received fewer pills at discharge.^5,8

While previous studies have examined aggregate opioid consumption in patients undergoing total joint arthroplasty in the first 30 days post-operatively, detailed information regarding when, why and how often opioids are consumed each day during this time period is lacking. The purpose of this study is to examine daily post-operative opioid frequency and quantity consumption for opioid-naïve patients undergoing total knee and hip arthroplasty (THA and TKA).

2. Methods

Level of Evidence: II. This study was an observational cohort study utilizing a patient journal for prospective data collection. Institutional review board approval was obtained and all participants signed a written informed consent form prior to participation. All patients scheduled for primary unilateral THA or TKA between June 2017 and November 2018 were approached for participation in the study. Patients were excluded if they were non-English speaking, had taken opioids at any time within 3 months prior to surgery, had a history of narcotic dependence or chronic opioid use, were scheduled for staged bilateral procedures, were expected to be discharged to a skilled nursing facility (SNF) or rehabilitation center based on preoperative discharge planning, or had a chronic pain condition which required long term pain medications. A total of 106 patients (45 THA, 61 TKA) agreed to participate and signed the informed consent form. Nine patients (5 THA, 4 TKA) were deemed ineligible post-consent because of an incorrectly coded surgery (revision, n = 1; staged bilateral procedures, n = 8), four patients (2 THA, 2 TKA) cancelled surgery, and 8 patients (4 THA, 4 TKA) withdrew their consent prior to surgery. Of the 85 remaining patients (34 THA, 51 TKA), five (1 THA, 4 TKA) were discharged to SNF instead of home. They were deemed ineligible and not provided a journal to complete. The remaining 80 patients (47 TKA, 33 THA) were provided journals at discharge.

During their regularly scheduled preoperative office visit, a member of the research team reviewed the instructions for completing the post-operative pain medication journal with each patient. For every dose of pain medication (narcotic and non-narcotic) taken each day, patients were asked to provide the type and dosage of medication, the quantity consumed, the pain score at the time of consumption, the time of day used, and whether and for how long the medication provided relief. Pain was rated using the Visual Analog Scale (VAS) from 1 to 10. A separate log was provided for each day following discharge, up to 30 days post-operatively. Patients were instructed to complete a medication log each day until they ceased using opioids for pain control. An example of a completed daily log is provided in Fig. 1.

Fig. 1.

An example of a completed medication journal daily entry.

All surgeries were performed by one of two fellowship-trained arthroplasty surgeons. Preoperatively, oral pain medications were given to all patients, which consisted of 1000 mg acetaminophen, 400 mg celecoxib, and 50 mg pregabalin. Most patients (73%) received general anesthesia. The proportion of patients receiving general anesthesia did not differ between TKA and THA groups (75% vs. 70%; p = 0.77). All TKA patients also received an adductor canal block preoperatively. All THA procedures were performed via the posterior approach using cementless implants. All TKA procedures were performed through the medial parapatellar approach using cemented, cruciate-retaining implants. An intraoperative periarticular injection, consisting of ropivacaine (300 mg), ketorolac (30 mg), epinephrine (1 mg), and clonidine (80 mcg) in 150 ml of normal saline, was administered for all patients prior to wound closure.

Post-operatively, patients received scheduled doses of 975 mg oral acetaminophen q8 hours and 15 mg intravenous ketorolac q6 hours. If additional pain control was needed during the hospital stay, TKA patients also received 5–10 mg oral oxycodone q6 hours, while THA patients received 50–100 mg oral tramadol q6 hours. If adequate pain control is not achieved with tramadol for THA patients, oxycodone is given. To manage pain following discharge, the majority of patients received a prescription for either oxycodone (5 mg) or tramadol (50 mg). For patients who could not tolerate oxycodone, hydromorphone (2 mg) or hydrocodone (5 mg) with acetaminophen was prescribed.

Prior to discharge, a member of the research team provided each patient with the pain medication journal, reviewed the instructions, and answered any questions. Following discharge, all patients were contacted once per week to monitor compliance with the log and answer any questions. If a patient had any questions or issues in between these calls, they were encouraged to contact a member of the study staff. Patients were asked to return the completed journals at their regularly scheduled post-operative clinic visit (five weeks post-surgery). If the patient had not ceased opioid use at this time, they were asked to continue filling out the journal and additional pages were provided. A member of the study staff continued to monitor these patients until the cessation of pain medication. Once completed, patients were asked to return the packets to a member of the study staff using a prepaid postage envelope provided to the patient at their post-operative clinic visit. In order to validate the number of opioid pills prescribed to each patient, specific opioid prescription information was obtained for each patient using their electronic medical record. Confirmation that a prescription was filled by the patient was assessed using our state's online prescription monitoring program.

Patient characteristics (age, sex, BMI, ASA physical status classification, Charleston Comorbidity Index (CCI)), along with relevant comorbidities (diabetes, hypertension, congestive heart failure, chronic lung conditions, depression/anxiety) were collected for each patient. Opioid usage data obtained from each patient journal included quantity and type of pain medications consumed, frequency of use, time of day used, intensity of pain associated with use, and if and for how long the medication provided relief. In order to standardize opioid data, dosages were converted to morphine equivalents (MEQ). Average inpatient and daily, weekly, and total post-discharge MEQ consumption was calculated, along with the pain score and hours of relief associated with consumption. The time of day of use was categorized into either day or nighttime pain. The number of filled prescriptions, quantity of unused pills, and the number of days from discharge that patients reported consuming opioids was determined for each patient.

Demographic and opioid data were summarized and presented separately for THA and TKA groups. Quantitative variables were described with a sample mean and standard deviation and examined for deviation from normality using the Shapiro-Wilk test. To compare data between TKA and THA groups, separate independent t-tests or Mann-Whitney U tests were used, as appropriate. Categorical variables were presented as percentages and examined using chi-square tests of independence. A multiple linear regression was calculated to predict total post-discharge MEQ consumption based on patient factors (age, CCI, BMI, type of surgery) and pre-discharge factors (average inpatient pain score, average inpatient MEQ, quantity of opioids prescribed at discharge).

3. Results

The study cohort consisted of 80 patients, with 33 undergoing THA (18 female, 15 male) and 47 undergoing TKA (25 females, 22 males). Three TKA patients were readmitted within 30 days of discharge, and their diaries were not included in the analysis. One patient had an infection and underwent a liner exchange, while the other two patients were admitted to the hospital with conditions not related to their operative joint. Completed diaries were returned for 55 patients (71%; 28 TKA, 27 THA). Of the 22 patients who did not return a completed journal, none reported taking opioids for pain management at their first post-operative visit. A check of the online opioid prescription program revealed that none of these patients received additional opioid prescriptions. Descriptive data for the TKA and THA patients who did complete the pain journal are presented in Table 1. There were no differences between the two groups for age (p = 0.15), sex distribution (p = 0.79), CCI (0.66), ASA score (0.76), and history of previous joint arthroplasty (p = 0.76); however, the TKA group did have a higher BMI (p = 0.002) and a longer length of stay (p = 0.048). A total of 13 patients (8 THA, 5 TKA) reported consuming no opioids following discharge after surgery. A check of the online prescription monitoring program confirmed that these patients did not fill a single opioid prescription during the post-operative period. All 13 patients reported that opiates were not required to control their pain. Adequate pain control was achieved in these patients using Tylenol.

Table 1.

Demographic and comorbidity data.

VARIABLE	ALL PATIENTS (N = 55)	THA PATIENTS (N = 27)	TKA PATIENTS (N = 28)	P-VALUE
Age	64.6 ± 8.1	63.0 ± 8.2	66.2 ± 7.8	0.150
Female Sex	53%	56%	50%	0.789
BMI	29.4 ± 5.6	27.1 ± 5.2	31.7 ± 5.1	0.002*
Length of Stay	2.2 ± 0.8	2.0 ± 0.7	2.4 ± 0.9	0.048*
Hypertension	40%	30%	50%	0.171
Cardiac Disease	13%	19%	7%	0.252
Lung Disease	2%	4%	0%	0.491
Diabetes	4%	0%	7%	0.491
Depression/Anxiety Disorder	17%	15%	18%	1.000
Charleston Comorbidity Indexa	2.0	2.0	2.0	0.658
ASA Score of 2	84%	82%	86%	0.586
History of Previous TJR	24%	26%	21%	0.758

*significant difference between TKA and THA patients.

^amedian value.

Overall, patients following total joint arthroplasty reported consuming less than half of their prescribed opioids (Table 2). Patients were prescribed an average of 452 MEQ post-operatively, yet reported consuming only 214 MEQ, leaving an average of 240 MEQ unused or the equivalent of 32 5-mg oxycodone pills. TKA patients both were prescribed (p = 0.005) and reported consuming (p = 0.005) a higher MEQ than THA patients. Overall opioid consumption averaged 123 MEQ during week 1 and 60 MEQ during week 2 post-discharge. Both daily average (p = 0.013) and weekly (p = 0.006) total MEQ consumption was greater for TKA patients compared to THA patients: however, consumption declined for both groups from post-operative week 1 to week 2 (Fig. 2 and Table 2). Of the total MEQ consumed post-operatively, THA patients took 41% of the total during days 1–3 and 77% of the total during week 1 post-discharge, while TKA patients took 38% of the total during days 1–3 and 70% of the total during week 1 after discharge. Patients reporting consuming opioids for an average of 14.5 days post-discharge, with only 35% of patients reporting consumption beyond 2-weeks post-discharge. TKA patients reported taking opioids for a longer period after surgery compared to THA patients (p = 0.004) (Table 2). Only 15% of THA patients were still taking opioids beyond 2 weeks post-discharge and no THA patient was still taking opioids at 3 weeks after surgery. For TKA patients, 50% reported opioid consumption after 2 weeks, which reduced to 30% after 3 weeks and 4% after 30 days’ post-discharge. The final patient reported ceasing opioid use at 31 days after discharge.

Table 2.

Opioid consumption data.

VARIABLE	ALL PATIENTS (N = 55)	THA PATIENTS (N = 27)	TKA PATIENTS (N = 28)	P-VALUE
Total MEQ consumed inpatient	74.3 ± 98.7	44.9 ± 47.8	102.7 ± 124.8	0.075
Number of opioid prescriptions	1.0 ± 0.7	0.8 ± 0.6	1.3 ± 0.8	0.010*
Prevalence of refill opioid prescriptions	24%	7%	39%	0.010*
Total MEQ prescribed	452.1 ± 259.5	296.7 ± 81.4	578.7 ± 345.4	0.005*
Total MEQ consumed post-discharge	214.1 ± 259.5	100.5 ± 112.2	323.6 ± 311.8	0.005*
During days 1–3 post-discharge	64.6 ± 72.9	34.8 ± 39.0	93.2 ± 86.2	0.005*
During week 1 post-discharge	123.1 ± 131.2	68.3 ± 75.9	176.0 ± 151.5	0.006*
During week 2 post-discharge	60.4 ± 83.5	27.4 ± 39.6	92.3 ± 101.4	0.006*
Average MEQ consumed per day post-discharge	14.7 ± 13.3	9.9 ± 9.6	19.3 ± 14.7	0.013*
During days 1–3 post-discharge	21.5 ± 24.3	11.6 ± 13.0	31.1 ± 28.7	0.005*
During week 1 post-discharge	19.1 ± 18.7	12.0 ± 12.5	26.1 ± 21.2	0.008*
During week 2 post-discharge	9.9 ± 11.8	5.0 ± 6.4	14.6 ± 13.9	0.004*
Last day of opioid use post-discharge	14.5 ± 7.5	10.9 ± 5.7	17.4 ± 7.6	0.004*

* significant difference between TKA and THA groups.

Fig. 2.

Daily morphine equivalent dose consumption for TKA and THA patients during the first two weeks post-operatively.

Overall, pain scores at the time of opioid consumption averaged 4.3/10 for all patients, with TKA patients reporting higher levels of pain compared to THA patients (p = 0.032) (Table 3). The intensity of pain reported at the time of opioid consumption was higher than pain scores reported when Tylenol was taken (4.3 vs. 3.7; p = 0.004) for all patients. For TKA patients, pain scores at time of opioid consumption decreased from week 1 to week 2 post-discharge (p = 0.014). THA patients reported a continued decrease in pain scores from days 1–3 to week 1 post-discharge (p = 0.012) and from week 1 to week 2 post-discharge (p < 0.001). Individual pain scores differed between TKA and THA patients during the first week post-discharge only (p = 0.045) (Table 3). Overall, patients reported experiencing 5 h of pain relief following opioid consumption, and hours of relief did not differ between TKA and THA patients (p = 0.28). A higher proportion of opioids were consumed at night (52%) compared to during the day (48%). Approximately two-thirds of all opioids consumed by THA patients were taken at night, compared to only 43% of opioids for TKA patients (p = 0.012).

Table 3.

Patient reported pain data.

VARIABLE	ALL PATIENTS (N = 55)	THA PATIENTS (N = 27)	TKA PATIENTS (N = 28)	P-VALUE
Average Pain Score at Opioid Consumption	4.3 ± 1.8	3.6 ± 2.0	4.8 ± 1.5	0.032*
During days 1–3 post-discharge	5.0 ± 2.0	4.5 ± 2.4	5.4 ± 1.6	0.200
During week 1 post-discharge	4.7 ± 1.9	4.1 ± 2.1	5.3 ± 1.5	0.045*
During week 2 post-discharge	3.9 ± 1.7	3.2 ± 1.6	4.3 ± 1.6	0.072
Hours of Relief from Opioids	4.9 ± 1.7	5.3 ± 2.0	4.6 ± 1.5	0.284
Average pain score during Inpatient	4.0 ± 1.6	3.5 ± 1.6	4.4 ± 1.5	0.031*
Average overall pain score	3.7 ± 1.8	3.4 ± 1.9	4.0 ± 1.6	0.252
Average pain score at Tylenol consumption	3.7 ± 1.8	3.4 ± 1.9	4.1 ± 1.7	0.252

* significant difference between TKA and THA groups.

Total post-discharge opioid consumption increased 42.6 MEQ for each one point increase in inpatient pain score, 1.1 MEQ for each one additional MEQ consumed inpatient, and 7.8 MEQ for each one pill increase in quantity of opioid pills prescribed at discharge, which is the equivalent of 5.7, 0.15, and 1.04 oxycodone 5 mg pills, respectively. In addition, TKA patients consumed 176.1 MEQ or 23.5 oxycodone 5 mg pills more than THA patients.

4. Discussion

Despite the widespread use of opioids for post-operative pain control following total joint replacement, data detailing actual patient usage is limited. Previous investigations have determined that the majority of patients consume fewer opioid pills than they are prescribed,^4,5 leaving large amounts of opioids at risk for diversion. Reducing prescription sizes at discharge may be associated with patients taking less opioids post-operatively⁵; however, to date, studies have only examined the first 30-days postoperatively in aggregate. In order to develop evidence-based standardized prescribing guidelines, determining opioid consumption during the entire post-operative period is needed. Therefore, we sought to examine daily post-operative opioid consumption for opioid-naïve patients undergoing THA and TKA procedures during the entire post-operative period. In addition, we investigated factors associated with increased opioid use. Our findings demonstrated that patients consumed less than half of their prescribed opioids, with over 70% of the total MEQ consumed during the first week following discharge. Total post-operative opioid consumption was associated with inpatient pain scores, inpatient opioid consumption, prescription size at discharge, and type of surgery.

Similar to previous studies, we found that the majority of patients do not require the total prescribed quantity of opioids to adequately control pain following TKA and THA procedures.^4,5 On average, patients in our study were prescribed 452 MEQ following surgery, with TKA patients receiving twice the number of opioids compared to THA patients. Overall, 53% of prescribed opioids went unused, with THA patients consuming only 1/3 of the quantity prescribed. Total MEQ consumed for our patients post-operatively (214 MEQ) was less than consumption rates reported in previous studies. Patient-reported opioid consumption during the first 30 days post-operatively in those studies averaged around 450 MEQ.^4,5 The difference in consumption rate may be due to our patients being prescribed less total opioids post-operatively, as prescribing a greater quantity of opioid pills has been associated with increased consumption.⁵ Our median initial prescription size was 40 pills, regardless of opioid type, and patients only received one type of opioid at discharge. Median prescription size for Huang et al. was 90 opioid pills⁴, and, while Hannon et al. had patients who received either 30 or 90 oxycodone pills, all patients in their study were also prescribed tramadol.⁵ As a result, while oxycodone usage was limited, the total MEQ was higher due to prolonged tramadol consumption. A critique of limiting the size of opioids prescribed at discharge is that it will result in patients requesting more refills. Despite prescribing fewer opioids at discharge, only 24% of our patients requested refill prescriptions, which is less than previously reported opioid refill rates.^2,3,8 Our data demonstrate that limiting the quantity of opioids prescribed following joint arthroplasty can be effective at reducing overall consumption without compromising pain control.

The need for opioids to manage post-operative pain lessened over time. Over 70% of all opioids consumed in our study were taken within the first week following surgery, regardless of procedure type, and average daily opioid consumption declined between week 1 and week 2 post-discharge. Our findings are similar to a previous study which reported that 75–80% of patients have stopped oxycodone usage by 3 weeks post-operatively; however, a greater number of their patients reported using tramadol for longer.⁵ The average number of days during which our patients reported using opioids to manage pain was 14, with consumption for TKA patients lasting an average of 7 days longer than THA patients. Only 50% of THA patients required opioids for pain management after the first week post-discharge, and only 15% reporting use greater than 2 weeks after discharge. Of those 15%, all reported taking only 1 pill a day as needed, with the majority consumed at night. For TKA patients, 50% reported using opioids at week 2 post-discharge, 30% at week 3, and only one patient required opioids greater than 30 days. The reduction in opioid usage over time was associated with a reduction in reported pain scores. The average pain score associated with opioid consumption decreased from week 1 to week 2 post-discharge for both TKA and THA patients. These data can be used to educate patients on post-operative pain expectations and develop effective opioid tapering guidelines.

Although our data show that opioids are not utilized for extended periods of time following joint replacement for most patients, there are some patients who do require opioids to manage their pain for a longer time. Factors that were associated with increased opioid consumption following joint replacement procedures included average inpatient pain score, total inpatient opioid consumption, quantity of opioids prescribed at discharge, and TKA surgery. Patients who experienced higher levels of pain and required more opioids to manage pain prior to discharge were more likely to consume a higher MEQ following discharge, approximately the equivalent of 6 more 5-mg oxycodone pills. In addition, average consumption for TKA patients was increased by 24 5-mg oxycodone pills compared to THA patients. Our data are similar to previous data that showed inpatient pain levels, TKA surgery, and number of pills prescribed at discharge as significant predictors of higher postoperative opioid consumption.⁵ These factors can be used to identify those patients at risk for needing more opioids to manage pain following joint replacement procedures and develop patient-specific pain management programs.

Our study is not without limitations. First, we relied on patients to accurately track their pain medication use following surgery, which puts the data at risk for recall bias. However, we attempted to mitigate this risk with regular follow-up calls to patients to review their journal. In addition, we confirmed any filled opioid prescriptions for each using our online prescription monitoring program, and no patient reported consuming more opioids than prescribed. We also recognize that consumption amounts may have been lower than normal because opioid use was being monitored and patients were concerned about the stigma associated with opioid use. Second, the medication diaries were incomplete or missing for 30% of our patients. All missing patients reported not taking opioids at their first post-operative visit, and we were able to confirm the lack of repeat prescriptions using the online prescription monitoring program. Despite this, we could not include them in the analysis because we could not provide specifics regarding their usage. We acknowledge that having their data may have resulted in differences in our consumption rates. Third, the possibility exists that our patients received prescriptions for opioids from prescribers outside of our state's integrated system. However, this risk is very minimal, as we were able to monitor filled prescriptions in 20 additional states. To further address this, we reviewed all pre-operative and post-operative clinical notes for each study patient to ensure that the patient reported cessation of opioids was consistent with prescription data. Finally, our sample size was not large; however, we feel that the breadth of data collected for each patient allowed us to capture an adequate representation of pain management for patients following TKA and THA.

5. Conclusions

Our findings revealed that patients consume, on average, less than half of their prescribed opioids following TKA and THA procedures, with over 70% of the total opioids consumed during the first week following discharge. The reduction in opioid consumption over the first two weeks post-operatively was associated with a reduction in reported pain scores. In addition, pre-discharge factors that were found to predict higher post-operative opioid consumption included inpatient pain scores, inpatient MEQ consumption, and undergoing TKA procedures. These data can be used to educate patients regarding pain expectations following surgery and develop evidence-based opioid tapering guidelines to effectively manage post-operative pain while limiting opioid use.

6. Author contributions statement

Authors Maureen K. Dwyer, Christopher M. Melnic, and Catharine Tumpowsky made significant contributions to research design, the acquisition, analysis or interpretation of data and drafting and revising the paper. Authors Tyler Humphrey, Shayan Hosseinzadeh, and Hany Bedair participated in visualization, drafting and revising the manuscript. All authors have read and approved the final manuscript.