Is the Preoperative Use of Antidepressants and Benzodiazepines Associated with Opioid and Other Analgesic Use After Hip and Knee Arthroplasty?

Abstract

Background

Mental health disorders can occur in patients with pain conditions, and there have been reports of an increased risk of persistent pain after THA and TKA among patients who have psychological distress. Persistent pain may result in the prolonged consumption of opioids and other analgesics, which may expose patients to adverse drug events and narcotic habituation or addiction. However, the degree to which preoperative use of antidepressants or benzodiazepines is associated with prolonged analgesic use after surgery is not well quantified.

Question/purposes

(1) Is the preoperative use of antidepressants or benzodiazepine medications associated with a greater postoperative use of opioids, NSAIDs, or acetaminophen? (2) Is the proportion of patients still using opioid analgesics 1 year after arthroplasty higher among patients who were taking antidepressants or benzodiazepine medications before surgery, after controlling for relevant confounding variables? (3) Does analgesic drug use decrease after surgery in patients with a history of antidepressant or benzodiazepine use? (4) Does the proportion of patients using antidepressants or benzodiazepines change after joint arthroplasty compared with before?

Methods

Of the 10,138 patients who underwent hip arthroplasty and the 9930 patients who underwent knee arthroplasty at Coxa Hospital for Joint Replacement, Tampere, Finland, between 2002 and 2013, those who had primary joint arthroplasty for primary osteoarthritis (64% [6502 of 10,138] of patients with hip surgery and 82% [8099 of 9930] who had knee surgery) were considered potentially eligible. After exclusion of another 8% (845 of 10,138) and 13% (1308 of 9930) of patients because they had revision or another joint arthroplasty within 2 years of the index surgery, 56% (5657 of 10,138) of patients with hip arthroplasty and 68% (6791 of 9930) of patients with knee arthroplasty were included in this retrospective registry study. Patients who filled prescriptions for antidepressants or benzodiazepines were identified from a nationwide drug prescription register, and information on the filled prescriptions for opioids (mild and strong), NSAIDs, and acetaminophen were extracted from the same database. For the analyses, subgroups were created according to the status of benzodiazepine and antidepressant use during the 6 months before surgery. First, the proportions of patients who used opioids and any analgesics (that is, opioids, NSAIDs, or acetaminophen) were calculated. Then, multivariable logistic regression adjusted with age, gender, joint, Charlson Comorbidity Index, BMI, laterality (unilateral/same-day bilateral), and preoperative analgesic use was performed to calculate odds ratios for any analgesic use and opioid use 1 year postoperatively. Additionally, the proportion of patients who used antidepressants and benzodiazepines was calculated for 2 years before and 2 years after surgery.

Results

At 1 year postoperatively, patients with a history of antidepressant or benzodiazepine use were more likely to fill prescriptions for any analgesics than were patients without a history of antidepressant or benzodiazepine use (adjusted odds ratios 1.9 [95% confidence interval 1.6 to 2.2]; p < 0.001 and 1.8 [95% CI 1.6 to 2.0]; p < 0.001, respectively). Similarly, patients with a history of antidepressant or benzodiazepine use were more likely to fill prescriptions for opioids than patients without a history of antidepressant or benzodiazepine use (adjusted ORs 2.1 [95% CI 1.7 to 2.7]; p < 0.001 and 2.0 [95% CI 1.6 to 2.4]; p < 0.001, respectively). Nevertheless, the proportion of patients who filled any analgesic prescription was smaller 1 year after surgery than preoperatively in patients with a history of antidepressant (42% [439 of 1038] versus 55% [568 of 1038]; p < 0.001) and/or benzodiazepine use (40% [801 of 2008] versus 55% [1098 of 2008]; p < 0.001). The proportion of patients who used antidepressants and/or benzodiazepines was essentially stable during the observation period.

Conclusion

Surgeons should be aware of the increased risk of prolonged opioid and other analgesic use after surgery among patients who were on preoperative antidepressant and/or benzodiazepine therapy, and they should have candid discussions with patients referred for elective joint arthroplasty about this possibility. Further studies are needed to identify the most effective methods to reduce prolonged postoperative opioid use among these patients.

Level of Evidence

Level III, therapeutic study.

Introduction

Hip and knee arthroplasties are highly effective surgical procedures for reducing pain in patients with late-stage osteoarthritis [18, 35]. This reduction in pain should eventually be reflected in reduced consumption of analgesic drugs. However, between 23% and 34% of patients who undergo hip or knee arthroplasty still use analgesic drugs (opioids, NSAIDs, or acetaminophen) 1 to 2 years after surgery, and between 5% and 16% still use opioids [11, 27, 36]. Moreover, although there may be other pain sites that patients and their physicians opt to treat with analgesics, between 10% and 20% of patients who undergo hip or knee arthroplasty continue to experience persistent postoperative pain, potentially explaining the increased consumption of opioids and other analgesics in this patient group [6, 10, 40]. Recently, attempts have been made to identify patients undergoing THA or TKA who do not benefit from surgery as well as expected and who continue to use analgesic drugs after surgery [4, 16, 22, 39]. Earlier, we found that analgesic drug consumption is reduced after hip and knee arthroplasty at the population level [36], and that obesity, a higher number of comorbidities, gender (women more than men), and preoperative use of analgesics were associated with the postoperative use of opioids, NSAIDs, or acetaminophen [38]. Previous studies have almost exclusively focused on opioid consumption, although psychiatric disorders, worse preoperative pain, and catastrophizing have also been associated with increased consumption of opioids after joint arthroplasty [4, 12, 22, 24, 25, 27, 32, 42, 43]. These factors have also been correlated with persistent pain [10, 28–30, 37, 45].

In some studies, patients undergoing THA or TKA who had preoperative depression or anxiety used opioids more frequently than patients without these disorders [4, 12, 14, 22, 25, 32, 34], whereas such an association has not been reported in other studies [22, 34, 42, 43]. Although NSAIDs and acetaminophen are used frequently in patients referred for joint arthroplasty [11, 20], previous studies have not analyzed whether the prior use of antidepressants and benzodiazepines is associated with the consumption of different types of analgesics (opioids, NSAIDs, or acetaminophen) after joint arthroplasty. It is also unclear whether opioid and overall analgesic use decreased after surgery in patients with a history of antidepressant or benzodiazepine use because previous studies [4, 12, 14, 22, 25, 32, 34, 42, 43] have focused only on factors associated with increased analgesic use, and no studies have reported whether analgesic drug use also decreased after surgery in patients with a history of antidepressant or benzodiazepine therapy. Moreover, comorbidities and potential adverse drug effects may influence the choice of analgesics (for example, starting opioids to avoid use of NSAIDS, or avoiding opioids in older patients). For these reasons, it is of interest to study both overall analgesic use as well as opioid use. It is also important to note that most previous studies have determined depression and anxiety from medical records based on diagnosis codes [3, 4, 25, 32, 34, 42, 43]. Information on antidepressant or benzodiazepine use may capture a higher proportion of patients at risk for prolonged opioid use compared with the use of diagnosis codes because hospital register data may be incomplete, and because it is not uncommon that these agents are used without a specific diagnosis of a psychiatric condition. Furthermore, benzodiazepines may be used for sleep disturbances related to chronic pain, and certain antidepressants are indicated for the treatment of chronic pain [19, 48]. Therefore, consumption of antidepressants and benzodiazepines might decrease after surgery [17, 26, 46]. We are aware of only one previous study on this subject, in which there was a minimal decrease in the proportion of patients who used benzodiazepines while the use of antidepressants remained stable, but this study included only hip arthroplasty patients [11].

To fill the gaps in previous studies, we asked: (1) Is the preoperative use of antidepressants or benzodiazepine medications associated with a greater postoperative use of opioids, NSAIDs, or acetaminophen? (2) Is the proportion of patients still using opioid analgesics 1 year after arthroplasty higher among patients who were taking antidepressants or benzodiazepine medications before surgery, after controlling for relevant confounding variables? (3) Does analgesic drug use decrease after surgery in patients with a history of antidepressant or benzodiazepine use? (4) Does the proportion of patients using antidepressants or benzodiazepines change after joint arthroplasty compared with before?

Patients and Methods

Study Design and Participants

We conducted a retrospective study of patients who underwent primary hip or knee arthroplasty for osteoarthritis between September 2002 and December 2013 in a single orthopaedic hospital in Finland (Coxa Hospital for Joint Replacement, Tampere, Finland). The study protocol has been described in detail [36]. The inclusion criteria were a primary operation and primary osteoarthritis as the indication for surgery. Other indications were excluded to maximize the homogeneity of our study group; for example, polyarticular involvement in patients with rheumatoid arthritis could confound analysis of the use of analgesic drugs. Only one procedure was included per patient (the first joint arthroplasty in the study period). Additionally, patients with revisions or primary arthroplasties of other joints during the observation period (2 years before or 2 years after the operation date of the index surgery) were excluded so that the potential perioperative peak in analgesic consumption related to this latter operation would not hamper the results related to the index surgery.

Of the 10,138 patients who underwent hip arthroplasty (with 13,802 hip arthroplasties) and 9930 patients who had knee arthroplasty (with 14,708 knee arthroplasties) during the observation periods, 64% (6502 of 10,138) of patients had primary hip arthroplasty and 82% (8099 of 9930) of patients had primary knee arthroplasty for primary osteoarthritis. A further 8% (845 of 10,138) and 13% (1308 of 9930) were excluded because of revision or another joint arthroplasty within 2 years of the index surgery, leaving 56% (5657 of 10,138) of hip and 68% (6791 of 9930) of knee arthroplasty patients for analysis (altogether 12,448 joint arthroplasty patients) (Fig. 1). Thanks to the use of nationwide register data, no patients were lost to follow-up.

Fig. 1.

The patients who were included in this study; OA = osteoarthritis. A color image accompanies the online version of this article.

Demographics

The mean age was 68 years (67 years for patients undergoing hip arthroplasty and 69 years for patients undergoing knee arthroplasty), and 61% (7550 of 12,448 patients) were women (53% [2971 of 5657] of patients with hip arthroplasty were women as were 67% [4579 of 6791] of those with knee arthroplasty). Among the 12,448 patients in the study, the most common comorbidities were hypertension (31% [3866]), cardiac disease (12% [1454]), and diabetes (9% [1168]). Preoperatively, antidepressants, benzodiazepines, or both were used by 5% (564), 12% (1534), and 4% (474) of patients, respectively (Table 1). Antidepressant use was more common in women and in patients who had a knee arthroplasty, unilateral joint replacement, higher Charlson Comorbidity Index score (CCI), diabetes, cardiac disease, psychotic disorder, neurodegenerative disease, pulmonary disease, history of malignancy, and epilepsy. Benzodiazepine use was associated with the same factors, as well as with hypertension and older age (see Supplementary Table 1; Supplemental Digital Content 1, http://links.lww.com/CORR/A561).

Table 1.

Table 1Demographic characteristics (n = 12,448 patients)

Factor	Value
Age in years, mean ± SD	68 ± 10
Women, % (n)	61 (7550)
Type of joint arthroplasty, % (n)
THA	45 (5657)
TKA	51 (6326)
Unicompartmental knee arthroplasty	4 (465)
Bilateral operation, % (n)	13 (1615)
BMI in kg/m2, mean ± SDa	29 ± 5
Charlson Comorbidity Index, % (n)b
0	71 (8887)
1	21 (2562)
>2	8 (999)
Diabetes, % (n)	9 (1168)
Cardiac disease, % (n)c	12 (1454)
Psychotic disorder, % (n)	2 (216)
Neurodegenerative disease (Alzheimer or Parkinson), % (n)	1 (181)
Pulmonary disease, % (n)	7 (931)
Hypertension, % (n)	31 (3866)
History of malignancy, % (n)	3 (424)
Epilepsy, % (n)	1 (132)
Preoperative antidepressant or benzodiazepine, % (n)
None	79 (9876)
Only antidepressant	5 (564)
Only benzodiazepine	12 (1534)
Both antidepressant and benzodiazepine	4 (474)

^aBMI missing on 13% (1644) of patients.

^bModified Charlson Comorbidity Index.

^cIncludes coronary artery disease, heart failure, and chronic arrythmia.

In the overall group, 3 months preoperatively, 42% (5281 of 12,448) of patients filled at least one type of analgesic prescription, most commonly NSAIDs (30% [3729 of 12,448]), followed by acetaminophen (12% [1484 of 12,448]), and opioids (11% [1339 of 12,448]). One year after surgery, the proportion of patients who filled at least one type of analgesic drug prescription had decreased to 25% (3157 of 12,448), and NSAIDs were still the most common group of analgesics used (15% [1927 of 12,448]), followed by acetaminophen (9% [1152 of 12,448]), and opioids (6% [777 of 12,448]).

Data Sources and Outcome Measures

Information on the drug use of these patients 2 years before and after surgery was gathered from Finland’s Drug Prescription Register, which is maintained by the Social Insurance Institution. Finland has a national health insurance scheme that covers all permanent residents, and the Drug Prescription Register contains information on all prescribed medications that have been dispensed from pharmacies in Finland. Information on the Anatomical Therapeutic Chemical code of the dispensed drugs, the number of units dispensed (tablets or patches), and the date of purchase regarding antidepressants (N06A), benzodiazepine derivatives (N05BA or N05CD), benzodiazepine-related drugs (N05CF), acetaminophen (N02BE01), NSAIDs (M01A), and opioids (N02A) was collected from the register. Benzodiazepine derivatives and benzodiazepine-related drugs were analyzed together and are referred to in this paper as benzodiazepines.

We analyzed analgesic drugs in two groups: opioids and any analgesic drugs (acetaminophen, NSAIDs, and opioids). Pooling all analgesic drugs was used to study overall need for analgesic drugs. Over-the-counter analgesic drugs were not recorded in this study. In Finland, over-the-counter analgesic drugs come in small packages of acetaminophen, ibuprofen, and ketoprofen, which are also dispensed by pharmacies and are relatively more expensive than those prescribed by a general practitioner or orthopaedic clinician. Although the study does not include data from the last few years, the guidelines for the pharmacological management of osteoarthritis pain have remained essentially similar through the study period and thereafter until 2019 [2, 9].

We divided patients into subgroups based on whether they had filled prescriptions for antidepressants or benzodiazepines 6 months preoperatively. Then, an exploratory analysis was made by calculating the proportions of patients who filled opioids and any analgesic prescriptions in these subgroups at 3-month intervals for a period of 2 years before and after surgery to illustrate the trends in drug use perioperatively. For the further statistical analyses, we analyzed analgesic use 3 months before surgery (referred to as preoperative drug use) and for 3 months at 9 to 12 months after surgery (referred to as 1 year postoperatively). We chose the period of 9 to 12 months after surgery to exclude any analgesic consumption related to the acute postoperative period because analgesic consumption was shown to stabilize at this timepoint in our previous analysis [36].

To examine our first and second research questions, we performed multivariable logistic regression for the probability of opioid and any analgesic use 3 months preoperatively and 1 year postoperatively. The multivariable model was adjusted for potentially confounding factors such as age, gender, joint, BMI (missing in 13% [1644 of 12,448] of patients), modified CCI, preoperative use of analgesics (categorical variable with three groups: opioid, NSAID or acetaminophen but no opioid, none), and whether the patient had same-day bilateral or unilateral surgery because these factors were associated with analgesic use in our previous analyses with a threshold of p < 0.05 [38]. Information on comorbidities for the CCI was extracted from the nationwide Special Reimbursement Register maintained by the Social Insurance Institution. A modified CCI score was calculated, as in an earlier study [38]. A score of 1 was assigned for heart failure, coronary artery disease, Types I or II diabetes, chronic asthma or other severe obstructive pulmonary disease, dementia, disseminated connective tissue diseases, rheumatoid arthritis, and other comparable conditions. A score of 2 was assigned for uremia resulting in dialysis, severe anemia in connection with chronic renal failure, leukemia and other malignant diseases of the blood and bone marrow (including malignant diseases of the lymphatic system), and cancer (including breast and prostate cancers, female genital tract cancer, and malignant neoplasms). Additionally, because epilepsy and psychotic disorders are not components of the CCI, we performed an additional analysis in which patients with these comorbidities were excluded, and the results were similar (this additional analysis was performed because a greater proportion of patients with epilepsy or psychosis used antidepressants and/or benzodiazepines preoperatively compared with patients without epilepsy or psychosis) (data not shown). The observed results were essentially the same among patients who had undergone hip or knee arthroplasty, and therefore, the results are presented together (for opioid use, see Supplementary Table 2; Supplemental Digital Content 2, http://links.lww.com/CORR/A562).

Table 2.

Multivariable adjusted ORs for analgesic consumption by patients with preoperative use of antidepressants or benzodiazepines

Factor	Preoperative (0-3 months)	Postoperative (9-12 months)
	Any analgesica Adjusted ORb, d	Any analgesica Adjusted ORc, d	Opioid Adjusted ORc, d
Antidepressants and benzodiazepines
None	1	1	1
Only antidepressant	1.6 (1.3-1.9)	1.9 (1.5-2.3)	2.2 (1.6-3.0)
Only benzodiazepine	1.9 (1.7-2.1)	1.8 (1.5-2.0)	1.9 (1.6-2.4)
Both antidepressant and benzodiazepine	2.2 (1.8-2.7)	2.4 (1.9-2.9)	2.8 (2.0-3.8)
Antidepressant (reference: no antidepressant)	1.7 (1.5-1.9)	1.9 (1.6-2.2)	2.1 (1.7-2.7)
Benzodiazepine (reference: no benzodiazepine)	1.9 (1.7-2.1)	1.8 (1.6-2.0)	2.0 (1.6-2.4)

^aAcetaminophen, NSAID, or opioid.

^bOR (95% CI) adjusted for age, gender, joint, Charlson Comorbidity Index score, BMI, and laterality (unilateral versus simultaneous bilateral).

^cOR (95% CI) adjusted for age, gender, joint, Charlson Comorbidity Index score, BMI, laterality (unilateral versus simultaneous bilateral), and preoperative use of opioids or other analgesics.

^dAll p values < 0.001.

For our third research question, we examined the exploratory trends of drug use in patients with a history of antidepressant or benzodiazepine use, and we compared the proportions of patients who used opioids and other analgesics preoperatively (0-3 months) and 1 year postoperatively. Finally, to answer question four, we descriptively reported the use of antidepressants and benzodiazepines preoperatively and postoperatively.

Statistical Analysis

The analyses were performed using SPSS Statistics, version 25 (IBM Corp). Parametric variables are presented as the mean and SD. We used a t-test to compare parametric variables and a chi-square test to compare categorical variables. We used the McNemar test to compare proportions in paired groups. Multivariable logistic regression analyses were used to calculate odds ratios with 95% confidence intervals for using analgesics. P values of < 0.05 were considered statistically significant.

Ethical Approval

Because this was a retrospective register study, no approval from our ethical board or consent from patients was required according to Finnish and EU legislation. The study was performed in accordance with the Declaration of Helsinki.

Results

Association Between Preoperative Antidepressants or Benzodiazepines and Postoperative Analgesic Use

After controlling for potentially confounding variables such as age, number of comorbidities (CCI), and preoperative use of analgesics, we found that patients with a history of antidepressant use were more likely to fill prescriptions for any analgesics than were patients without a history of antidepressant use (adjusted odds ratio 1.9 [95% confidence interval 1.6 to 2.2]; p < 0.001) (Table 2). Similarly, patients with a history of benzodiazepine use were more likely to fill prescriptions for any analgesics than were patients without a history of benzodiazepine use (adjusted OR 1.8 [95% CI 1.6 to 2.0]; p < 0.001) (Table 2). Moreover, the odds were the highest in patients with a preoperative use of both antidepressants and benzodiazepines (Table 2). At 1 year postoperatively, analgesic prescriptions of all types were filled by 22% (2137 of 9876) of patients who had no history of preoperative use of antidepressants or benzodiazepines compared with 39% (219 of 564), 38% (581 of 1534), and 46% (220 of 474) of patients with prior use of antidepressants, benzodiazepines, or both, respectively (p < 0.001) (Fig. 2). The proportion of patients who continued to use analgesics or started to use analgesics 1 year postoperatively (those who did not use analgesics preoperatively but used them 1 year postoperatively) was higher in patients with preoperative use of antidepressants or benzodiazepines (Table 3).

Fig. 2.

The proportions (with 95% CI) of patients with any analgesic drug use according to the preoperative use of antidepressants and benzodiazepines, in 3-month intervals. A color image accompanies the online version of this article.

Table 3.

Continuation of analgesic use according to preoperative antidepressant and benzodiazepine use

		Use of any analgesics
		Continueda	Stopped after surgeryb	Started after surgeryc	No use pre- or postoperativelyd	p value
Antidepressant preoperatively	Yes (n = 1038)	30 (314)	24 (254)	12 (125)	33 (345)
	No (n = 11,410)	16 (1778)	26 (2935)	8 (940)	50 (5757)
						< 0.001
Benzodiazepine preoperatively	Yes (n = 2008)	28 (570)	26 (528)	12 (231)	34 (679)
	No (n = 10,440)	15 (1522)	26 (2661)	8 (834)	52 (5423)
						< 0.001

		Use of opioids
		Continued	Stopped after surgery	Started after surgery	No use pre- or postoperatively	p value
Antidepressant preoperatively	Yes (n = 1038)	9 (92)	11 (112)	6 (62)	74 (772)
	No (n = 11,410)	2 (270)	8 (865)	3 (353)	87 (9922)
						< 0.001
Benzodiazepine preoperatively	Yes (n = 2008)	6 (126)	11 (216)	6 (124)	77 (1542)
	No (n = 10,440)	2 (236)	7 (761)	3 (291)	88 (9152)
						< 0.001

Data presented as % (n).

^aRedeemed medications 3 months preoperatively and 1 year postoperatively.

^bRedeemed medications 3 months preoperatively but not 1 year postoperatively.

^cDid not redeem medications 3 months preoperatively but redeemed 1 year postoperatively.

^dDid not redeem medications 3 months preoperatively or 1 year postoperatively.

Association Between Preoperative Antidepressants or Benzodiazepines and Persistent Opioid Use After Surgery

In the multivariable logistic regression analysis, we found that patients with a history of antidepressant use were more likely to fill prescriptions for opioids than were patients without a history of antidepressant use (adjusted OR 2.1 [95% CI 1.7 to 2.7]; p < 0.001). Patients with a history of benzodiazepine use were also more likely to fill prescriptions for opioids than were patients without a history of benzodiazepine use (adjusted OR 2.0 [95% CI 1.6 to 2.4]; p < 0.001). At 1 year postoperatively, opioid prescriptions were filled by 5% (455 of 9876) of patients who had no history of preoperative use of antidepressants or benzodiazepines, compared with 13% (72 of 564) of patients with a history of only antidepressant use, 11% (168 of 1534) of patients with a history of only benzodiazepine use, and 17% (82 of 474) of patients with a history of both antidepressant and benzodiazepine use (p < 0.001) (Fig. 3). The proportion of patients who continued to use opioids (those who filled opioid prescriptions preoperatively and postoperatively) or started to use opioids (those who did not use opioids preoperatively but used them 1 year postoperatively) was higher in patients with preoperative use of antidepressants or benzodiazepines (Table 3). The results were essentially the same among patients who had undergone hip or knee arthroplasty (for opioid use, see Supplementary Table 2; Supplemental Digital Content 2, http://links.lww.com/CORR/A562).

Fig. 3.

The proportions (with 95% CI) of patients with opioid use according to the preoperative use of antidepressants and benzodiazepines, in 3-month intervals. A color image accompanies the online version of this article.

Decrease in Analgesic Use After Surgery in Patients Taking Antidepressants or Benzodiazepines

Among the 1038 patients with a history of using antidepressants, the proportion of patients who filled any analgesic prescriptions was smaller 1 year after surgery (42% [439] of patients), than preoperatively (55% [568] of patients; p < 0.001) (Fig. 2), and the proportion who filled a prescription for opioids was smaller 1 year after surgery (15% [154] of patients), than preoperatively (20% [204] of patients; p < 0.001) (Fig. 3). Among the 2008 patients with a history of using benzodiazepines, the proportion of patients who filled any analgesics was smaller 1 year after surgery (40% [801]) than preoperatively (55% [1098]; p < 0.001) (Fig. 2), and the proportion who filled a prescription for opioids was smaller 1 year after surgery (12% [250]) than preoperatively (17% [342]; p < 0.001) (Fig. 3).

Change in Antidepressant or Benzodiazepine Use After Arthroplasty

The proportion of patients using antidepressants did not change from the preoperative period (7% [844 of 12,448]) to 1 year after surgery (7% [845 of 12,448]; p > 0.99) (Fig. 4). The proportion of patients who filled benzodiazepines was slightly smaller 1 year postoperatively compared with the preoperative period (12% [1501 of 12,448] versus 13% [1566 of 12,448]; p = 0.04) (Fig. 4).

Fig. 4.

The proportions (with 95% CI) of patients with antidepressant or benzodiazepine use, in 3-month intervals.

Altogether, the proportion of patients who filled antidepressant prescriptions was essentially stable over the study period (Fig. 4). The proportion of patients using benzodiazepines increased slightly preoperatively, peaked immediately after surgery, and decreased thereafter to a lower level than that seen at 3 months preoperatively (Fig. 4). The results were similar between patients treated with hip arthroplasty and those treated with knee arthroplasty.

Discussion

Persistent postoperative pain affects between 10% and 20% of patients who undergo hip or knee arthroplasty, which may result in the prolonged consumption of opioids and other analgesics in this patient group [6, 10, 40]. Because of the ongoing opioid epidemic, attempts have been made to identify especially those patients undergoing THA or TKA who continue to use opioids after surgery, but disagreement persists as to whether preoperative depression and anxiety are associated with persistent opioid use after surgery [4, 12, 14, 22, 25, 32, 34, 42, 43]. In this large registry study, we found that patients with a history of antidepressant or benzodiazepine use were more likely to fill prescriptions for opioids and other analgesics 1 year after surgery than were patients without a history of antidepressant or benzodiazepine use. The increased possibility of prolonged opioid and other analgesic use should be considered and discussed before performing joint arthroplasty on these patients.

Limitations

The most important limitation of this registry study is that we were not able to control for all potentially confounding variables, although major confounders, such as the preoperative use of analgesics and comorbidities were included in the adjusted regression model. Another limitation is that we pooled all analgesic drugs (acetaminophen, NSAIDs, and opioids) to study the overall need for analgesic drugs. Although this approach allowed us to capture a greater proportion of patients who had pain after surgery, there was most likely a great difference in pain intensity between patients who irregularly used acetaminophen and patients who regularly used strong opioids. Additionally, the risks related to prolonged opioid use are at a completely different level compared with acetaminophen use. Therefore, we analyzed opioid use separately. Some patients may have entered drug treatment/detoxification programs, but it is unlikely that this confounded the results. Opioid abuse has been very rare in Finland; even across the country there would have been very few, if any, patients having joint arthroplasty during drug treatment/detoxification programs, although it may be more common in patients on antidepressants or benzodiazepines. In addition, we did not analyze the amounts of drugs used. Therefore, the results only present whether there is a need for opioids or other analgesic drugs. We were unable to assess the indications for the prescribed analgesics. Therefore, we were not able to determine whether patients filled analgesic prescriptions because of pain in the operated joint or because of pain in other sites, although we tried to reduce confounding by excluding patients with revisions or other joint arthroplasties during the study period and by excluding patients with indications other than primary osteoarthritis. But even if some patients had arthritis in other joints (besides the one operated on for primary osteoarthritis), this should not undermine our results considering the prolonged usage of analgesics in patients with history of antidepressant and benzodiazepine use. Inability to assess medication adherence is another limitation, and we were not able to find out whether the filled prescriptions were actually taken by the patient and when. The data that we used were relatively old (they span the period from 2002 to 2013). However, we do not believe that this would confound the association with preoperative antidepressant/benzodiazepine and analgesic use, or the other questions studied.

Another limitation is that over-the-counter drug use was not analyzed, which may have underestimated the actual use of acetaminophen and NSAIDs. It is unlikely, however, that this would have substantially confounded the drug use patterns or the risk factors analyzed. Furthermore, not all postoperative complications were analyzed. However, the major complications leading to revision surgery of the index joint were excluded. Pain variables, such as the intensity of preoperative and postoperative pain, the prevalence of persistent pain, or pain catastrophizing, were not analyzed in this study. It was not possible to determine the severity of mental health problems and whether the medication used was sufficient. The generalizability of the results may be limited because all surgical procedures were performed in the same hospital. However, although there may be some variation in the indications for joint replacement and prescription practices in other centers, we believe the results apply to osteoarthritis patients in other countries because the national guidelines for treatment of osteoarthritis in Finland are similar to other Western countries.

Association Between Preoperative Antidepressants or Benzodiazepines and Postoperative Analgesic Use

The coexistence of depression and chronic pain often aggravates the severity of both disorders [41], and up to 85% of patients with chronic pain are affected by severe depression [1, 21, 41]. In our study, there was a clear association between a history of antidepressant or benzodiazepine use and the use of any of the studied analgesics (opioids, NSAIDs, and acetaminophen), both preoperatively and postoperatively. Only a few studies have analyzed the association between depression and analgesics other than opioids, and a greater use of NSAIDs has been found after hip and knee arthroplasty [42, 43], even though there is a synergistic risk of gastrointestinal bleeding between serotonin reuptake inhibitors and NSAIDs [7]. In the present study, a history of benzodiazepine use was associated with the postoperative use of not only opioids but also NSAIDs and acetaminophen, which to our knowledge has not been described before. Benzodiazepine use has been associated with pain severity and catastrophizing in patients with chronic pain, and this may be one explanation for this finding [15, 33].

Association Between Preoperative Antidepressants or Benzodiazepines and Persistent Opioid Use After Surgery

Although opioids are effective drugs for treating chronic pain, major risks, such as dependence and addiction, are associated with their use; in general, opioids are not a good choice for treating arthritic pain in most patients. Emerging evidence suggests that opioids do not provide benefit when compared with NSAIDs to manage noncancer pain (such as arthritis pain), but opioids are associated with a higher risk for adverse events, such as dependence and addiction [13]. Additionally, preoperative opioid use is associated with worse patient outcomes after total joint arthroplasty [23]. The latest Osteoarthritis Research Society International (OARSI) guidelines for the nonsurgical management of osteoarthritis pain do not recommend opioid use in patients with osteoarthritis [2]. However, opioid prescriptions have been increasing so that recently around 24% to 60% of patients are being prescribed opioids before total joint arthroplasty [4, 5, 27, 32]. Some studies have found that patients undergoing hip and knee arthroplasty who have depression preoperatively use opioids more frequently after surgery than patients without depression [4, 12, 22, 25, 32], whereas no such association has been reported in other studies [34, 43]. Similarly, in some earlier studies, anxiety and the preoperative use of benzodiazepines was associated with greater use of opioids after hip and knee arthroplasty [4, 12, 14, 25, 32, 34, 43], whereas such an association has not been reported in other studies [22, 42].

In our study, after adjusting for possible confounders such as preoperative analgesic use and comorbidities, there was a clear association between a history of antidepressant or benzodiazepine use and the prolonged use of opioids after surgery. Moreover, the probability of prolonged opioid use was greatest in patients with a preoperative use of both antidepressants and benzodiazepines. Most earlier studies identified depression and anxiety from preoperative medical records based on registered diagnosis codes, which limits the analysis to specific psychiatric conditions; however, benzodiazepines, in particular, are used more widely as anxiolytics and sleep medicines without a specific psychiatric diagnosis. Our approach may have helped capture a higher proportion of patients at risk for prolonged opioid use compared with the use of diagnosis codes. Based on the results of our study, surgeons should be aware of the increased possibility of prolonged opioid and other analgesic use among patients who have surgery while on preoperative antidepressant and/or benzodiazepine therapy. Further studies are needed to find ways to reduce the risk for prolonged opioid consumption in these patients; these might, for example, explore the effects of more intensive preoperative management of depression and anxiety, alternative pain management techniques, more intensive monitoring, and patient counseling.

Decrease in Analgesic Use After Surgery in Patients Taking Antidepressants or Benzodiazepines

The proportion of patients who used opioids and other analgesics decreased after surgery to a lower level than that observed preoperatively. This was also true for patients undergoing antidepressant or benzodiazepine therapy, although the proportion of patients who used opioids and other analgesics remained higher in these patients than in patients without a history of antidepressant or benzodiazepine use. Surprisingly, we did not find any previous studies on this subject because earlier studies only analyzed depression and anxiety as risk factors for increased analgesic use. Some previous studies have found that patients with depression or anxiety are often satisfied after joint arthroplasty, and that surgery has positive effects on the pain levels of patients with depression [8, 26]. Our study adds to these previous studies: The consumption of opioids and other analgesics is reduced after surgery in patients with a history of antidepressant or benzodiazepine therapy. However, these patients should be informed of the greater possibility of prolonged opioid and other analgesic use after surgery, and depression and anxiety should be managed as effectively as possible before elective joint arthroplasty surgery.

Change in Antidepressant or Benzodiazepine Use After Arthroplasty

In accordance with the findings of a study on patients undergoing hip arthroplasty [11], the proportion of patients using antidepressants remained essentially stable throughout the study period, except for a minor decrease immediately after surgery. Following evidence-based recommendations [19], serotonin-noradrenaline reuptake inhibitors (duloxetine and venlafaxine) and tricyclic antidepressants may be used in the treatment of chronic neuropathic pain. In Finland, as well as in many other countries, these drugs are also used in patients with mixed pain patterns if the treating clinician believes there are both nociceptive and neuropathic components of pain involved. However, our findings suggest that these patients most probably used antidepressants for indications other than chronic pain related to osteoarthritis of the index joint because the proportion of patients who filled antidepressant prescriptions remained essentially stable throughout the study period. The proportion of patients using benzodiazepines increased before surgery, peaked immediately postoperatively, and then decreased. However, the amplitude of the changes in the proportion of patients using benzodiazepines was minimal and was similar to that reported in a previous study [11]. There was a slight increase in the long-term proportions during the observation period from 2 years preoperatively to 2 years postoperatively. The preoperative increase of benzodiazepine use in patients waiting for elective joint arthroplasty may be related to the link between pain and sleep or between pain and anxiety [11, 44, 47]. Generally, the preoperative use of benzodiazepines in our study was in line with an earlier study [11], but the preoperative use of antidepressants (9%) was slightly higher than in a previous study [27] and in a Finnish population study [31]. Neither the use of antidepressants nor benzodiazepines changed markedly in the long-term after joint arthroplasty. Based on present results, the proportion of patients who use antidepressants or benzodiazepines should not be expected to change very much after joint arthroplasty.

Conclusion

In this large, register-based study with comprehensive drug prescription data, we found that the preoperative use of antidepressants and benzodiazepines was associated with more sustained postoperative use of opioids and other analgesics. Sustained opioid and other analgesic use was more common in patients who filled both antidepressants and benzodiazepines preoperatively than it was in patients who used only one of those two drug classes before surgery, and more common than in patients who filled neither antidepressants nor benzodiazepines. Based on the results of this study, surgeons should be aware of the increased possibility of prolonged opioid and other analgesic use among patients undergoing preoperative antidepressant and/or benzodiazepine therapy, and they should have candid discussions with patients referred for elective joint arthroplasty about this possibility. Further studies are needed to identify the most effective methods to reduce prolonged opioid consumption among these patients.