Research

JAMA Surgery | Original Investigation

Drug-Free Interventions to Reduce Pain or Opioid

Consumption After Total Knee Arthroplasty
A Systematic Review and Meta-analysis
Dario Tedesco, MD; Davide Gori, MD; Karishma R. Desai, PhD; Steven Asch, MD, MPH; Ian R. Carroll, MD; Catherine Curtin, MD;
Kathryn M. McDonald, MM; Maria P. Fantini, MD; Tina Hernandez-Boussard, PhD

Supplemental content
IMPORTANCE There is increased interest in nonpharmacological treatments to reduce pain
after total knee arthroplasty. Yet, little consensus supports the effectiveness of these
interventions.

OBJECTIVE To systematically review and meta-analyze evidence of nonpharmacological

interventions for postoperative pain management after total knee arthroplasty.

DATA SOURCES Database searches of MEDLINE (PubMed), EMBASE (OVID), Cochrane Central
Register of Controlled Trials (CENTRAL), Cochrane Database of Systematic Reviews, Web of
Science (ISI database), Physiotherapy Evidence (PEDRO) database, and ClinicalTrials.gov for
the period between January 1946 and April 2016.

STUDY SELECTION Randomized clinical trials comparing nonpharmacological interventions

with other interventions in combination with standard care were included.

DATA EXTRACTION AND SYNTHESIS Two reviewers independently extracted the data from
selected articles using a standardized form and assessed the risk of bias. A random-effects
model was used for the analyses.

MAIN OUTCOMES AND MEASURES Postoperative pain and consumption of opioids and
analgesics.

RESULTS Of 5509 studies, 39 randomized clinical trials were included in the meta-analysis
(2391 patients). The most commonly performed interventions included continuous passive
motion, preoperative exercise, cryotherapy, electrotherapy, and acupuncture.
Moderate-certainty evidence showed that electrotherapy reduced the use of opioids (mean
difference, −3.50; 95% CI, −5.90 to −1.10 morphine equivalents in milligrams per kilogram per
48 hours; P = .004; I2 = 17%) and that acupuncture delayed opioid use (mean difference,
46.17; 95% CI, 20.84 to 71.50 minutes to the first patient-controlled analgesia; P < .001;
I2 = 19%). There was low-certainty evidence that acupuncture improved pain (mean
difference, −1.14; 95% CI, −1.90 to −0.38 on a visual analog scale at 2 days; P = .003; I2 = 0%).
Very low-certainty evidence showed that cryotherapy was associated with a reduction in
opioid consumption (mean difference, −0.13; 95% CI, −0.26 to −0.01 morphine equivalents in
milligrams per kilogram per 48 hours; P = .03; I2 = 86%) and in pain improvement (mean
difference, −0.51; 95% CI, −1.00 to −0.02 on the visual analog scale; P < .05; I2 = 62%).
Low-certainty or very low-certainty evidence showed that continuous passive motion and
preoperative exercise had no pain improvement and reduction in opioid consumption: for
continuous passive motion, the mean differences were −0.05 (95% CI, −0.35 to 0.25) on the
visual analog scale (P = .74; I2 = 52%) and 6.58 (95% CI, −6.33 to 19.49) opioid consumption
at 1 and 2 weeks (P = .32, I2 = 87%), and for preoperative exercise, the mean difference was
−0.14 (95% CI, −1.11 to 0.84) on the Western Ontario and McMaster Universities Arthritis
Index Scale (P = .78, I2 = 65%). Author Affiliations: Author
affiliations are listed at the end of this
CONCLUSIONS AND RELEVANCE In this meta-analysis, electrotherapy and acupuncture after article.
total knee arthroplasty were associated with reduced and delayed opioid consumption. Corresponding Author: Tina
Hernandez-Boussard, PhD,
Department of Medicine, Stanford
JAMA Surg. doi:10.1001/jamasurg.2017.2872 University, 1265 Welch Rd, Stanford,
Published online August 16, 2017. CA 94305 (boussard@stanford.edu).

(Reprinted) E1
© 2017 American Medical Association. All rights reserved.

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

 Research Original Investigation Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty

T
here are 234 million major surgical procedures per-
formed every year worldwide, and most patients expe- Key Points
rience moderate to severe postoperative pain.1,2 Inad-
Question Which of the nonpharmacological interventions used
equate postoperative pain management has profound acute for postoperative pain after total knee arthroplasty are effective?
effects, including immune system suppression, decreased mo-
Findings In a systematic review of 5509 studies, 39 randomized
bility that increases deep vein thrombosis and pulmonary em-
clinical trials were included in a meta-analysis (2391 patients) and
bolism rates, myocardial infarction, and pneumonia.3 Long-
demonstrated moderate-certainty evidence that electrotherapy
term influences of poor pain management include transition to and acupuncture reduce or delay opioid consumption, but there is
chronic pain and prolonged narcotic consumption,4 which can low certainty or very low certainty that they improve pain.
result in opioid dependence, an epidemic in the United States.5 Continuous passive motion and preoperative exercise do not
First-line therapies to treat postoperative pain are pharma- improve pain or reduce opioid consumption (low certainty or very
cological, including anesthetics, opioids, and acetaminophen.6,7 low certainty), and cryotherapy reduces opioid consumption but
does not improve pain (very low certainty).
Recently, nonpharmacological approaches to pain management
aimed at reducing the use of prescription medications have Meaning After total knee arthroplasty, electrotherapy and
increased.8 Physiotherapy is effective in treating postoperative acupuncture were associated with reduced and delayed opioid
pain and quality-of-life improvement and is standard treatment.9 consumption.
However, other commonly used interventions for pain manage-
ment have conflicting evidence on their effectiveness.10,11 As opi-
oid addiction becomes a national priority,5 the importance of EMBASE (OVID), Cochrane Central Register of Controlled Trials
using effective nonpharmacological strategies for postoperative (CENTRAL), Cochrane Database of Systematic Reviews, Web
pain is now a top scientific priority.12 of Science (ISI database), Physiotherapy Evidence (PEDRO) da-
Total knee arthroplasty (TKA) is one of the most fre- tabase, and clinicaltrials.gov. We scanned reference lists of se-
quently performed surgical procedures worldwide. It is used for lected reviews, original articles, and textbooks to find addi-
patients with advanced knee osteoarthritis, and the goals of sur- tional articles. We conducted a gray literature search for other
gery are to decrease pain, restore mobility and function, and im- documents and hand searches of key conference proceed-
prove health-related quality of life.13 Total knee arthroplasty is ings, journals, professional organizations’ websites, national
associated with intense postoperative pain, and many pa- joint replacement registries, and guideline clearing houses.
tients report moderate to severe postoperative pain past the an- Snowball technique was applied to the search strategy.20
ticipated recovery period.14 Therefore, many nonpharmaceu-
tical therapies are performed in this population.10,11,15,16 Ensuring Search Criteria
effective therapies for postoperative pain management is an Because our aim was to be as comprehensive as possible in the
important part of TKA care.17 systematic review, we did not place time or publication sta-
We undertook a systematic review and meta-analysis to tus limits to the search except for restriction to the English lan-
evaluate the effectiveness of commonly used drug-free inter- guage. Two Chinese studies with English abstracts were con-
ventions for pain management after TKA. We gathered evi- sidered; however, translation resources were not available to
dence from randomized clinical trials (RCTs) on postopera- include them. We used the following search string in each da-
tive pain as measured by established pain metrics and reduced tabase: (postoperative pain* OR postoperative pain OR post-
analgesic consumption, including opioids and nonsteroidal operative pain) AND (total knee* or total knee arthroplasty OR
anti-inflammatory drugs (NSAIDs). This comprehensive analy- total knee replacement OR TKA). Asterisks are used to trun-
sis of nonpharmaceutical pain management therapies can in- cate words, so that every desinence after the asterisks will
form practice and identify effective pain management regi- be searched. To achieve the highest sensitivity, we used a com-
mens that could also potentially reduce the prescribing of bination of keywords and indexed terms (eg, PubMed Medi-
opioids after surgery. cal Subject Headings).

Patients, Interventions, Comparators, and Outcomes

Our primary search objective addressed the PICO (patients,
Methods interventions, comparators, and outcomes) question. These
We conducted a systematic review and meta-analysis in ac- targets included (P) patients undergoing primary TKA, (I)
cord with the Preferred Reporting Items for Systematic Re- nonpharmacological treatments for pain management (plus
views and Meta-Analyses (PRISMA).18 The protocol was reg- usual analgesic therapy), (C) other nonpharmacological inter-
istered in the International Prospective Register of Systematic vention or no intervention (plus usual analgesic therapy),
Reviews (PROSPERO).19 and (O) postoperative pain relief, and opioid and NSAID
consumption.
Search Strategy
Databases Study Selection
In an academic medical setting, we searched electronic data- Design
bases to identify relevant studies for the period between Janu- Postoperative pain management is generally layered, includ-
ary 1946 and April 2016, including MEDLINE (PubMed), ing pharmacological and nonpharmacological interventions.

E2 JAMA Surgery Published online August 16, 2017 (Reprinted) jamasurgery.com

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

 Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty Original Investigation Research

Hence, we selected studies comparing nonpharmacological in- nism of acupuncture analgesia is its effect on the central ner-
terventions with routine pharmacological treatment either with vous system, particularly a short-term and long-term effect on
other nonpharmacological approaches or with only routine μ-opioid receptors, and consequent regulation of neurotrans-
pharmacological treatments. We restricted our meta-analysis mitters and hormones.25,26
to RCTs in which patients were 18 years or older and had elec-
tive primary surgical procedures that included all forms of Study Quality Assessment
fixation (cemented, hybrid, or cementless), surgical ap- Two of us (D.T. and D.G.) independently assessed the risk of
proaches (medial, lateral, parapatellar, or minimally invasive), bias of included studies using the parameters defined by the
and types of prostheses (constrained, semiconstrained, or mo- Cochrane Handbook for Systematic Reviews of Interventions
bile platform). criteria.27 Disagreement was resolved through discussion and
consensus between the reviewers. Based on the information
Outcomes of Interest provided from included studies, each item was recorded as low
Three of us (D.T., D.G., and K.R.D.) independently screened all risk of bias, high risk of bias, or unclear (lack of information
identified articles by scanning abstracts or portions of the text or unknown risk of bias).
to determine if they met the inclusion criteria. Any disagree- Two of us (D.T. and D.G.) independently assessed the
ments were resolved through discussion and consensus be- quality of the body of evidence for the different outcomes
tween the reviewers. Postoperative pain relief was defined as considered through the Grades of Recommendation, Assess-
the mean difference in scores on the visual analog scale (VAS) ment, Development, and Evaluation (GRADE) approach, a
or the Western Ontario and McMaster Universities Arthritis In- validated and widely implemented tool to rate the quality of
dex Scale (WOMAC). Opioid and other analgesic consump- scientific evidence.28 According to the GRADE approach, we
tion was evaluated in terms of the mean difference in con- assessed 5 domains, grading the strength of evidence for
sumption of morphine equivalents in milligrams per kilogram each outcome.
per 48 hours, while other analgesic consumption was evalu-
ated as the mean difference in the number of tablets per day. Data Analysis and Synthesis
Time to first request for analgesia (patient-controlled analge- Three of us (D.T., D.G., and K.R.D.) independently extracted
sia [PCA]) in the acupuncture group was defined as minutes the data from included articles. Key information was gath-
from the end of the intervention to the first PCA. ered systematically using a standardized form. These vari-
Information about quality of life was not systematically ables included country, year of publication, number of par-
provided in the studies. Therefore, it was not included. ticipants, intervention, age, sex, study design, duration of
intervention, outcome time points, statistical method, post-
Intervention operative pain, opioid or analgesic consumption, and sum-
We restricted our focus to commonly studied postoperative mary of the results.
pain interventions. These included continuous passive mo- For the pain scores, we standardized the results to a single
tion (CPM), preoperative exercise, cryotherapy, electro- scale by converting outcomes reported on a numerical rating
therapy, and acupuncture. scale to a 10-point VAS. Where possible, the results were ex-
Continuous passive motion consists of using an external tracted manually from the published figures. Data in other
machine to provide regular movement to the knee using a pre- forms (ie, median, interquartile range, and mean [95% CI]) were
determined range of motion (ROM). Theoretically, the re- converted to means (SDs) according to the Cochrane Hand-
peated movements help increase ROM, while simultane- book for Systematic Reviews of Interventions.27 If data (eg, SDs
ously improving pain.21 and SEs) were not presented in the original article, correspond-
Preoperative exercise (or prehabilitation) involves ing authors were contacted to acquire the missing data, al-
sessions performed by the patients in the weeks preceding though no responses were received. We also normalized data
surgery. This regimen enables them to cope better with the for pain relief and analgesic consumption, opioid and other an-
physical stress associated with the surgical procedure and aids algesic consumption, and time before the first analgesic treat-
postoperative rehabilitation efforts.22 ment. Specifically, all data on opioid consumption were con-
Cryotherapy is based on applying cold to the surgical site verted to milligrams per kilogram per 48 hours, other analgesic
either through ice bags or cooled water to minimize tissue consumption was converted to the number of tablets per day,
trauma. The theory is that application of cooler substances re- and time before the first analgesic treatment was converted
duces intra-articular temperatures, which interferes with the to minutes.
conduction of nerve signals and reduces local blood flow. These We examined the evidence tables for clinical (partici-
changes lead to decreased swelling and perceived pain.23 pants, interventions, controls, outcomes, and measurement
Electrotherapy (based on electrophysical agents) aims to tools) and methodological heterogeneity to determine whether
reduce pain and improve function through an energy trans- the studies were similar enough to perform a meta-analysis.29
fer to the body. These modalities include transcutaneous elec- Where appropriate to pool the results, we used weighted mean
trical nerve stimulation and pulsed electromagnetic fields.24 differences for continuous data using the same measurement
Acupuncture is a form of traditional Chinese medicine that scales and standardized mean differences for continuous
requires the insertion of needles at specific points on the body outcomes using different scales. We pooled both sets of sum-
to alleviate pain and other ailments. The plausible mecha- mary statistics using the inverse variance method, which

jamasurgery.com (Reprinted) JAMA Surgery Published online August 16, 2017 E3

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

 Research Original Investigation Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty

included studies from different time points, and we con- ticularly in those testing the effectiveness of CPM, cryotherapy,
ducted sensitivity analyses by single time point. and electrotherapy RCTs.32,35,37,40,41,43,53,55,57,58,63,65,68 There
We tested statistical heterogeneity to determine if it was was also high risk of bias due to improper or absent random se-
appropriate to combine the studies for meta-analysis. We ex- quencing methods in 8 studies.33,34,42,44,45,47,57,59 Last, a study65
amined heterogeneity graphically using forest plots and sta- in the electrotherapy group showed high risk of bias for incom-
tistically by calculating the I2 statistic, which describes the per- plete outcome data. We conducted sensitivity subgroup analy-
centage of the variability in effect estimates that is due to ses for all the outcomes considered, classifying for sequence
heterogeneity rather than sampling error (chance). We con- generation and allocation concealment availability, and no sig-
sidered an I2 statistic greater than 50% to be substantially nificant differences were shown (eFigures 10, 11, 12, 13, 14, 15,
heterogeneous. According to the Cochrane Handbook for Sys- 16, 17, 18, and 19 in the Supplement). The GRADE quality of
tematic Reviews of Interventions,27 in cases where the num- evidence certainty level of evidence assessment is reported in
ber of studies was less than 5 or studies were substantially detail below in the Assessed Outcomes and Evidence Synthe-
heterogeneous, we used a random-effects model. We calcu- sis subsection, in Table 2, and in eTable 2 in the Supplement.
lated the random-effects estimates for the corresponding sta-
tistics using the method by DerSimonian and Laird.30 Forest Publication Bias
plots were created to display effect estimates with 95% CIs for To address publication bias, we created funnel plots for all
individual trials and pooled results. For all data analysis, we analyses. No asymmetric patterns were seen (eFigures 24, 25,
used a software program (RevMan, version 5.3; The Cochrane 26, 27, 28, and 29 in the Supplement).
Collaboration).
Interventions
The key findings of the meta-analysis are summarized in
Table 2 for 2 types of pain scales and for 3 types of analgesic
Results outcomes. Figure 1 and Figure 2 show the meta-analyses that
Search Findings reported statistically significant results.
Supplementary information is provided in eTables 1, 2, and 3
in the Supplement and in eFigures 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, Assessed Outcomes and Evidence Synthesis
12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, Pain Relief and Analgesic Consumption
30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, and 42 in the Supple- We found that the quality of evidence was of low certainty or
ment. Our search yielded 5509 studies, of which 120 (112 from very low certainty for pain improvement in all examined in-
selection and 8 added by hand and snowball searching) were terventions (Table 2). Meta-analysis of 2 pain relief studies (189
appropriate for further assessment. Of the 77 RCTs we read in patients) suggested a significant improvement in experimen-
extenso, we extracted the data from 39 RCTs for our meta- tal groups vs controls with electrotherapy, with mean differ-
analysis (eFigure 1 in the Supplement). Included studies were ences of −1.95 (95% CI, −2.68 to −1.22; P < .001; I2 = 17%) on
published between 1991 and 2015. the VAS at 1 month, −2.34 (95% CI, −4.49 to −0.19; P = .03;
I2 = 94%) on the VAS at 2 months, and −2.60 (95% CI, −5.12 to
Study Characteristics −0.08; P = .04; I2 = 83%) on the VAS at 6 months (Figure 1A).
A pooled total of 2391 patients were examined in the RCTs Meta-analysis of 3 studies (230 patients) suggested a signifi-
(Table 1). We categorized the 39 RCTs based on 2 outcomes cant improvement in experimental groups vs controls with acu-
(pain relief and analgesic consumption, including different puncture, with a mean difference of −1.14 (95% CI, −1.90 to
measures and types) and 5 interventions, including 18 stud- −0.38; P = .003; I2 = 0%) on the VAS at 6 months (Figure 1B).
ies in the CPM group (14 on pain and 5 on analgesics), 3 stud- Meta-analysis of 8 studies (1383 patients) showed a mean
ies in the preoperative exercise group (all on pain), 12 studies difference with cryotherapy of −0.51 (95% CI, −1.00 to −0.02;
in the cryotherapy group (8 on pain and 10 on analgesics), 4 P < .05; I2 = 62%), but all subgroup analyses showed no sta-
studies in the electrotherapy group (2 on pain and 2 on anal- tistically significant mean differences (eFigure 2 in the Supple-
gesics), and 4 studies in the acupuncture group (2 on pain and ment). Meta-analysis of 9 studies (1025 patients) suggested no
3 on analgesics). One study48 recurred in 3 different catego- significant improvement in experimental groups vs controls
ries owing to multiple comparison groups within the article. with CPM (mean differences, −0.05; 95% CI, −0.35 to 0.25;
For the studies that did not provide sufficient data, we at- P = .74; I2 = 52% on the VAS at 1 week and 6 months and −0.20;
tempted to contact authors but received no response. 95% CI, −0.62 to 0.23; P = .54; I2 = 0% on the CPM WOMAC at
6 weeks and 6 months) (eFigure 3 and eFigure 4 in the Supple-
Quality Assessment ment) or with preoperative exercise (mean difference, −0.14;
All studies were assessed for the risk of bias (eTable 1 in the 95% CI, −1.11 to 0.84; P = .78; I2 = 65% on the WOMAC at 6 and
Supplement). The methodological heterogeneity reflects the 12 weeks) (eFigure 5 in the Supplement).
different range of interventions we examined. We identified that To address possible overestimation that could originate
the highest bias in studies was due to improper or absent mask- from the study design (ie, pain as a primary or secondary out-
ing during the study (31 of 39 RCTs). In 2 studies54,58 on cryo- come), we conducted sensitivity subgroup analyses. No sig-
therapy, masking was adequately achieved. Studies also showed nificant differences were found (eFigures 20, 21, 22, and 23 in
high risk of bias for selective outcome reporting in 13 cases, par- the Supplement).

E4 JAMA Surgery Published online August 16, 2017 (Reprinted) jamasurgery.com

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

 Table 1. Characteristics of Randomized Clinical Trials of Nonpharmacological Postoperative Pain Management After Total Knee Arthroplasty
No. of Participants in Age, Mean (SD) or Median
Source Country Per-Protocol Analysis Intervention (Range), y Female, %
CPM

jamasurgery.com
Beaupré et al,31 2001 Canada 120 T1: CPM T1: 68 (9) T1: 30
T2: Slider board T2: 68 (9) T2: 50
C: Standard exercise C: 69 (8) C: 52.5
Bennett et al,32 2005 Australia 147 T1: Standard CPM T1: 70.7 T1: 72.3
T2: Early flexion CPM T2: 71.4 T2: 64.6
C: No CPM C: 71.7 C: 67.3
Bruun-Olsen et al,33 2009 Norway 63 T: CPM plus active exercise 68 (10) T: 73
C: Active exercise 71 (10) C: 67
Chen et al,34 2013 Taiwan 107 T: CPM plus basic rehabilitation T: 69.3 (6.8) NA
C: Basic rehabilitation C: 69.5 (8.2) NA
Colwell and Morris35 1992 United States 22 T: CPM plus standard rehabilitation T: 73 T: 67
C: Immobilization in a posterior splint plus standard C: 74 C: 70
rehabilitation
Denis et al,36 2006 Canada 81 T1: Low-intensity CPM plus conventional physical therapy T1: 69.6 (6.7) T1: 61.5
T2: High-intensity CPM plus conventional physical therapy T2: 68.4 (7.4) T2: 46.4
C: Conventional physical therapy C: 67.1 (7.6) C: 51.9
Harms and Engstrom,37 1991 United Kingdom 113 T: CPM plus standardized exercise T1: 69 (9) T1: 78
C: Standardized exercise C: 71 (10) C: 93
Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty

Kim et al,38 2009 South Korea 100 T: Regular PROME plus standard exercise 67.9 (53-83) 100
C: Standard exercise (no PROME) NA NA

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

Lenssen et al,39 2003 The Netherlands 40 T: CPM plus physical therapy T: 65 (9) T: 71
C: Physical therapy C: 66 (10) C: 63
Lenssen et al,40 2008 The Netherlands 60 T: CPM plus physical therapy T: 68.9 (6.1) T: 60
C: Physical therapy C: 67.5 (8.9) C: 70

© 2017 American Medical Association. All rights reserved.

MacDonald et al,41 2000 Canada 120 T1: CPM with ROM 0°-50° NA NA
T2: CPM with ROM 70°-110° NA NA
C: No CPM NA NA
Maniar et al,42 2012 India 86 T1: 1 d of CPM plus conventional physical therapy T1: 66.8 T1: 89
T2: 3 d of CPM plus conventional physical therapy T2: 66 T2: 87
C: Conventional physical therapy C: 67.4 C: 92
May et al,43 1999 United States 19 T: CPM plus physical therapy T: 73 (4) T: 67
C: Lower limb mobility board plus physical therapy C: 66 (9) C: 71
McInnes et al,44 1992 United States 93 T: CPM plus conventional rehabilitation T: 65.7 (1.6) T: 65
C: Conventional rehabilitation C: 70.2 (1.3) C: 64

(continued)

(Reprinted) JAMA Surgery Published online August 16, 2017

Original Investigation Research

E5
 E6
Table 1. Characteristics of Randomized Clinical Trials of Nonpharmacological Postoperative Pain Management After Total Knee Arthroplasty (continued)
No. of Participants in Age, Mean (SD) or Median
Source Country Per-Protocol Analysis Intervention (Range), y Female, %
Montgomery and Eliasson,45 1996 United Kingdom 60 T: CPM T: 74 (5) T: 86
C: Active physical therapy C: 76 (6) C: 75
Pope et al,46 1997 Australia 53 T1: CPM 0°-40° ROM plus physical therapy T1: 72.5 T1: 64.7
T2: CPM 0°-70° ROM plus physical therapy T2: 72.7 T2: 50
C: Physical therapy C: 72.2 C: 69.6
Research Original Investigation

Sahin et al,47 2006 Turkey 31 T: CPM plus physical therapy T: 61 (6.0) T: 86

C: Physical therapy C: 61.6 (7.5) C: 86
Walker et al,48 1991 United States 22 T: CPM T: 72.7 (61-83) NA
C: No CPM C: 73.6 (65-82)
Preoperative Exercise
Calatayud et al,49 2016 Spain 44 T: 8 wk of exercise training program 3 d per wk before surgery T: 66.8 (4.8) 84
C: No intervention NA NA

JAMA Surgery Published online August 16, 2017 (Reprinted)

Gstoettner et al,50 2011 Austria 35 T: 6 wk of preoperative proprioceptive training program T: 72.8 (65-78) NA
C: No intervention C: 66.9 (61-75) NA
McKay et al,51 2012 Canada 22 T: 6 wk of prehabilitation exercise training program T: 60.58 (8.05) T: 66.67
C: No intervention C: 63.5 (4.93) C: 50
Cryotherapy
Albrecht et al,52 1997 Germany 98 T1: Intermittent ice blocks T1: 69.8 NA
T2: Continuous cold therapy T2: 71.9 NA
C: Standard postoperative care plus CPM C: 73.8 NA
Gibbons et al,53 2001 United Kingdom 60 T: Cold compression dressing T: 70 T: 63

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

C: Robert Jones modified bandage C: 71 C: 53
Ivey et al,54 1994 United States 88 T1: Hot ice thermal pads at 10°C T1: 64.5 T1: 57
T2: Hot ice thermal pads at 15.6°C T2: 64.2 T2: 63
C: Hot ice thermal pads at 21.1°C C: 66.9 C: 73
Kullenberg et al,55 2006 Sweden 83 T: Cold compression applied with the Cryo Cuff (Aircast T: 68.1 (6) T: 74

© 2017 American Medical Association. All rights reserved.

Incorporated)
C: Standard care C: 68.9 (6.8) C: 70
Levy and Marmar,56 1993 United States 80 T: Cold compression applied with the Cryo Cuff (Aircast T: 74 T: 82.5
Incorporated)
C: No intervention C: 73 C: 80
Morsi,57 2002 Egypt 60 T: Cooling device NA NA
C: Standard care NA NA

(continued)

jamasurgery.com
Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty
 Table 1. Characteristics of Randomized Clinical Trials of Nonpharmacological Postoperative Pain Management After Total Knee Arthroplasty (continued)
No. of Participants in Age, Mean (SD) or Median
Source Country Per-Protocol Analysis Intervention (Range), y Female, %
Radkowski et al,58 2007 United States 64 T: Cryotherapy at 7.2°C continued until discharge plus elastic T: 63.7 T: 54

jamasurgery.com
wrap
C: Cryotherapy at 23.9°C continued until discharge plus elastic C: 66.9 C: 36
wrap
Smith et al,59 2002 Australia 84 T: Cold therapy T: 72.1 (7.8) T: 54.8
C: Compression bandage C: 72 (7.1) C: 45.2
Su et al,60 2012 United States 187 T: Cryopneumatic device NA NA
C: Ice and static compression NA NA
Thienpont,61 2014 Belgium 100 T: Continuous cooling at 11°C T: 67.5 (10.5) T: 70
C: Cold packs (conserved at −17°C) C: 68.5 (10) C: 80
Webb et al,62 1998 United Kingdom 31 T: Cold compression with the Cryo Cuff (Aircast Incorporated) T: 69.0 NA
C: Standard care C: 70.9 NA
Walker et al,48 1991 United States 30 Continuous cooling pad plus CPM T: 75.0 (58-87) NA
CPM C: 70.0 (56-82) NA
Electrotherapy
Adravanti et al,63 2014 Italy 26 T: Pulsed electromagnetic fields plus standard rehabilitation T: 66 (13) T: 62.5
C: Standard rehabilitation C: 73 (5) C: 53
Borckardt et al,64 2013 United States 40 T: Transcranial direct current stimulation NA NA
C: Sham transcranial direct current stimulation NA NA
Moretti et al,65 2012 Italy 30 T: Stimulation with pulsed electromagnetic fields 4 h per d for T: 70.5 (8.1) NA
Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty

60 d plus kinesitherapy
C: Kinesitherapy C: 70.0 (10.6) NA

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

Walker et al,48 1991 United States 48 T1: TENS plus CPM T: 69.8 (61-69) NA
T2: Subthreshold TENS plus CPM T2: 73.9 (55-84) NA
C: CPM C: 73.9 (60-86) NA
Acupuncture

© 2017 American Medical Association. All rights reserved.

Chen et al,66 2015 Taiwan 60 T: Auricular acupuncture under general anesthesia T: 68.9 (9.0) T: 90
C: Sham auricular acupuncture under general anesthesia C: 69.0 (8.6) C: 73
Mikashima et al,67 2012 Japan 80 T: Acupuncture treatment plus standard rehabilitation T: 72 (7) T: 75
C: Standard rehabilitation C: 73 (5) C: 70
Tsang et al,68 2007 Hong Kong 30 T: Acupuncture plus standard rehabilitation T: 70.6 (5.8) T: 80
C: Sham acupuncture plus standard rehabilitation C: 66.1 (7.5) C: 80
Tzeng et al,69 2015 Taiwan 47 T1: Electroacupuncture T1: 70.1 (6.9) T1: 82
T2: Sham electroacupuncture T2: 69.6 (5.6) T2: 75
C: No intervention C: 71.4 (7.3) C: 71

Abbreviations: C, control group; CPM, continuous passive motion; NA, not available; PROME, passive ROM exercise; ROM, range of motion; T, treatment group; TENS, transcutaneous electrical nerve stimulation.

(Reprinted) JAMA Surgery Published online August 16, 2017

Original Investigation Research

E7
 Research Original Investigation Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty

Table 2. Main Findings of the Meta-analysis of Nonpharmacological Postoperative Pain Management After
Total Knee Arthroplastya
No. of I2
No. of Partici- Heteroge-
Variable Studies pants Effect Estimate (95% CI) neity, % GRADE
Pain Relief on the VAS
CPM 9 1025 −0.05 (−0.35 to 0.25) 52
1 wk 8 575 −0.27 (−0.70 to 0.16) 35
2 wk 2 145 −0.81 (−3.30 to 1.68) 88 Very low
3 mo 2 170 0.50 (−0.30 to 1.29) 54
6 mo 2 135 0.15 (−0.04 to 0.35) 0
Cryotherapy 8 1382 −0.51 (−1.00 to −0.02) 62
Day 1 after surgery 7 529 −0.21 (−0.89 to 0.48) 40
Very low
Day 2 after surgery 5 422 −1.00 (−2.01 to 0.02) 59
Day 3 after surgery 6 431 −0.44 (−1.37 to 0.49) 75
Electrotherapy 2 189 −2.11 (−2.74 to −1.47) 80
1 mo 2 63 −1.95 (−2.68 to −1.22) 17
Very low
2 mo 2 63 −2.34 (−4.49 to −0.19) 94
6 mo 2 63 −2.60 (−5.12 to −0.08) 83
Acupuncture 3 230 −0.66 (−1.29 to −0.03) 69
2d 2 90 −1.14 (−1.90 to −0.38) 0 Low
8d 2 140 −0.37 (−1.04 to 0.30) 72
Pain Relief on the WOMAC
CPM 5 578 0.03 (−0.19 to 0.24) 0
6 wk 2 168 −0.66 (−2.12 to 0.81) 60
Low
3 mo 3 242 0.05 (−0.22 to 0.32) 0
6 mo 2 168 −0.34 (−1.35 to 0.68) 0
Preoperative exercise 3 132 −0.14 (−1.11 to 0.84) 65
6 wk 2 60 0.34 (−0.32 to 0.99) 0 Low
12 wk 2 72 −0.78 (−1.63 to 0.07) 8
Opioid Consumptionb
CPM 5 313 6.58 (−6.33 to 19.49) 87
1 wk 3 178 11.12 (−12.21 to 34.44) 80 Very low Abbreviations: CPM, continuous
2 wk 2 135 −3.78 (−7.67 to 0.11) 8 passive motion; GRADE, Grades of
Cryotherapy within 48 h 7 468 −0.13 (−0.26 to −0.01) 86 Recommendation, Assessment,
Development, and Evaluation;
Cryotherapy vs nothing 2 61 −0.18 (−0.30 to −0.06) 0 Very low NA, not available;
Cryotherapy vs compression 5 407 −0.12 (−0.28 to 0.04) 90 NSAID, nonsteroidal
Electrotherapy within 48 h 2 99 −3.50 (−5.90 to −1.10) 17 Moderate anti-inflammatory drug;
PCA, patient-controlled analgesia;
Acupuncture within 48 h 2 123 −0.71 (−1.44 to 0.02) 64
VAS, visual analog scale (range, 0-10);
Acupuncture vs sham 2 90 −0.92 (−2.19 to 0.35) 79 WOMAC, Western Ontario and
Low
acupuncture McMaster Universities Arthritis Index
Acupuncture vs nothing 1 33 −0.40 (−1.05 to 0.25) NA Scale (range, 0-20).
a
NSAID Consumption The inverse variance method was
Cryotherapy within 48 h 3 363 −0.75 (−1.63 to 0.12) 95 used for the statistical analyses.
b
Units are the mean use of morphine
Cryotherapy vs nothing 1 60 −1.90 (−2.25 to −1.55) NA Very low
equivalents in milligrams at 1 and 2
Cryotherapy vs compression 2 303 −0.31 (−0.55 to −0.07) 0 weeks for CPM, the number of
Time to First PCAc tablets per 48 hours for
cryotherapy, and morphine
Acupuncture 2 124 46.17 (20.84 to 71.50) 19
equivalents in milligrams per
Acupuncture vs sham 2 91 34.58 (−12.61 to 81.77) 53 kilogram per 48 hours for all other
Moderate
acupuncture
variables under this heading.
Acupuncture vs nothing 1 33 57.90 (16.52 to 99.28) NA c
Units are minutes after surgery.

Opioid and Other Analgesic Consumption analysis of 7 studies (468 patients) showed very low-certainty
Meta-analysis of 2 studies (99 patients) showed moderate- reduction in opioid consumption for cryotherapy (mean differ-
certainty reduction in opioid consumption for electrotherapy ence, −0.13; 95% CI, −0.26 to −0.01 opioids in milligrams per
(mean difference, −3.50; 95% CI, −5.90 to −1.10 opioids in mil- hour; P = .03; I2 = 86%) (Figure 2B). Meta-analysis of 3 studies
ligrams per hour; P = .004; I 2 = 17%) (Figure 2A). Meta- (363 patients) showed very low-certainty nonreduction in

E8 JAMA Surgery Published online August 16, 2017 (Reprinted) jamasurgery.com

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

 Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty Original Investigation Research

Figure 1. Pain Relief and Analgesic Consumption

A Electrotherapy

Electrotherapy Control
Mean Difference Favors Favors Weight,
Source Mean (SD) Total Mean (SD) Total (95% CI) Electrotherapy Control %
Electrotherapy at 1 mo
Adravanti et al,63 2014 2.5 (1.4) 17 4.2 (0.7) 16 –1.70 (–2.45 to –0.95) 18.7
Moretti et al,65 2012 2.4 (1.6) 15 4.9 (1.8) 15 –2.50 (–3.72 to –1.28) 13.1
Subtotal (95% CI) 32 31 –1.95 (–2.68 to –1.22) 31.8
Heterogeneity: τ2 = 0.05; χ21 = 1.20 (P =.27); I2 = 17%
Test for overall effect: z = 5.26 (P <.001)
Electrotherapy at 2 mo
Adravanti et al,63 2014 1.4 (0.5) 16 2.7 (0.4) 17 –1.30 (–1.61 to –0.99) 23.7
Moretti et al,65 2012 1.1 (1.0) 15 4.6 (1.8) 15 –3.50 (–4.54 to –2.46) 15.0
Subtotal (95% CI) 31 32 –2.34 (–4.49 to –0.19) 38.7
Heterogeneity: τ2 = 2.27; χ21 = 15.73 (P < .001); I2 = 94%
Test for overall effect: z = 2.13 (P =.03)
Electrotherapy at 6 mo
Adravanti et al,63 2014 0.4 (0.2) 16 1.9 (0.9) 17 –1.50 (–1.94 to –1.06) 22.4
Moretti et al,65 2012 1.5 (2.8) 15 5.6 (2.9) 15 –4.10 (–6.14 to –2.06) 7.1
Subtotal (95% CI) 31 32 –2.60 (–5.12 to –0.08) 29.5
Heterogeneity: τ2 = 2.81; χ21 = 5.96 (P =.01); I2 = 83%
Test for overall effect: z = 2.02 (P =.04)
Total (95% CI) 94 95 –2.11 (–2.74 to –1.47) 100.0
Heterogeneity: τ2 = 0.42; χ25 = 24.45 (P < .001); I2 = 80%
Test for overall effect: z = 6.46 (P <.001)
Test for subgroup differences: χ22 = 0.33 (P =.85); I2 = 0%

–8 –6 –4 –2 0 2 4 6 8
Mean Difference (95% CI)

B Acupuncture
Acupuncture Control
Mean Difference Favors Favors Weight,
Source Mean (SD) Total Mean (SD) Total (95% CI) Acupuncture Control %
Acupuncture at 2 d
Chen et al,66 2015 5.7 (1.7) 15 6.5 (1.5) 15 –0.80 (–1.95 to 0.35) 17.0
Tsang et al,68 2007 4.6 (1.5) 30 6.0 (2.4) 30 –1.40 (–2.41 to –0.39) 19.4
Subtotal (95% CI) 45 45 –1.14 (–1.90 to –0.38) 36.3
Heterogeneity: τ2 = 0.00; χ21 = 0.59 (P = .44); I2 = 0%
Test for overall effect: z = 2.94 (P =.003)
Acupuncture at 8 d
Chen et al,66 2015 4.5 (1.3) 30 5.3 (1.4) 30 –0.80 (–1.48 to –0.12) 26.7
Mikashima et al,67 2012 5.8 (0.5) 40 5.9 (0.6) 40 –0.10 (–0.34 to 0.14) 37.0
Subtotal (95% CI) 70 70 –0.37 (–1.04 to 0.30) 63.7
Heterogeneity: τ2 = 0.18; χ21 = 3.58 (P =.06); I2 = 72%
Test for overall effect: z = 1.09 (P =.27)
Total (95% CI) 115 115 –0.66 (–1.29 to –0.03) 100.0
Heterogeneity: τ2 = 0.26; χ23 = 9.79 (P =.02); I2 = 69%
Test for overall effect: z = 2.05 (P =.04)
Test for subgroup differences: χ21 = 2.18 (P =.14); I2 = 54.2%

–8 –6 –4 –2 0 2 4 6 8
Mean Difference (95% CI)

Shown are individual and pooled weighted mean differences in pain measured days and 8 days after surgery.66-68 A random-effects model was used to pool
with a visual analog scale using the inverse variance method. A, The mean the data.
differences at 1, 2, and 6 months after surgery.63,65 B, The mean differences at 2

nonsteroidal antiinflammatory drug (NSAID) consumption for with 60 patients) and −0.31 (95% CI, −0.55 to −0.07 tablet per
cryotherapy (mean difference, −0.75; 95% CI, −1.63 to 0.12 tab- day; P = .01; I2 = 0%) for cryotherapy vs compression (2 stud-
lets per day; P = .09; I2 = 95%). Nevertheless, subgroup analy- ies, with 303 patients) (eFigure 6 in the Supplement). Acupunc-
ses showed a significant reduction in NSAID consumption, with ture (2 studies, with 123 patients) and CPM (5 studies, with 313
mean differences of −1.90 (95% CI, −2.25 to −1.55 tablets per day; patients) showed no significant differences between experi-
P < .01; I2 = not applicable) for cryotherapy vs nothing (1 study, mental groups and controls for amount of opioid consumed

jamasurgery.com (Reprinted) JAMA Surgery Published online August 16, 2017 E9

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

 Research Original Investigation Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty

Figure 2. Opioid and Other Analgesic Consumption

A Electrotherapy

Electrotherapy Control
Mean Difference Favors Favors Weight,
Source Mean (SD) Total Mean (SD) Total (95% CI) Electrotherapy Control %
Borckardt et al,64 2013 6.3 (5.6) 20 12.3 (6.6) 19 –6.00 (–9.85 to –2.15) 32.3
Walker et al,48 1991 6.6 (5.7) 18 8.7 (5.0) 12 –2.10 (–5.96 to 1.76) 32.1
Walker et al,48 1991 6.2 (4.9) 18 8.7 (5.0) 12 –2.50 (–6.12 to 1.12) 35.7
Total (95% CI) 56 43 –3.50 (–5.90 to –1.10) 100.0
Heterogeneity: τ2 = 0.77; χ22 = 2.42 (P = .30); I2 = 17%
Test for overall effect: z = 2.86 (P =.004)

–10 –8 –6 –4 –2 0 2 4 6 8 10
Mean Difference (95% CI)

B Cryotherapy
Cryotherapy Control
Mean Difference Favors Favors Weight,
Source Mean (SD) Total Mean (SD) Total (95% CI) Cryotherapy Control %
Cryotherapy vs nothing
Walker et al,48 1991 0.622 (0.169) 15 0.844 (0.293) 15 –0.22 (–0.39 to –0.05) 13.2
Webb et al,62 1998 0.570 (0.230) 15 0.710 (0.230) 16 –0.14 (–0.30 to 0.02) 13.6
Subtotal (95% CI) 30 31 –0.18 (–0.30 to –0.06) 26.8
Heterogeneity: τ2 = 0.00; χ21 = 0.46 (P =.50); I2 = 0%
Test for overall effect: z = 2.98 (P =.003)
Cryotherapy vs compression
Gibbons et al,53 2001 0.650 (0.310) 30 0.600 (0.310) 30 0.05 (–0.11 to 0.21) 13.8
Kullenberg et al,55 2006 0.370 (0.110) 43 0.430 (0.050) 40 –0.06 (–0.10 to –0.02) 17.6
Levy and Marmar et al,56 1993 0.530 (0.200) 40 0.960 (0.300) 40 –0.43 (–0.54 to –0.32) 15.5
Smith et al,59 2002 0.320 (0.290) 44 0.420 (0.310) 40 –0.10 (–0.23 to 0.03) 14.9
Thienpont et al,61 2014 0.792 (0.563) 50 0.802 (0.542) 50 –0.01 (–0.23 to 0.21) 11.4
Subtotal (95% CI) 207 200 –0.12 (–0.28 to 0.04) 73.2
Heterogeneity: τ2 = 0.03; χ24 = 41.77 (P <.001); I2 = 90%
Test for overall effect: z = 1.44 (P =.15)
Subtotal (95% CI) 237 231 –0.13 (–0.26 to –0.01) 100.0
Heterogeneity: τ2 = 0.02; χ26 = 44.38 (P <.001); I2 = 86%
Test for overall effect: z = 2.17 (P =.03)
Test for subgroup differences: χ12 = 0.38 (P =.54); I2 = 0%

–1.0 –0.5 0 0.5 1.0

Mean Difference (95% CI)

Shown are individual and pooled weighted mean differences in opioid consumption within 48 hours after surgery (morphine equivalents in milligrams per kilogram
per 48 hours) using the inverse variance method. A, Electrotherapy.48,64 B, Cryotherapy.48,53,55,56,59,61,62 A random-effects model was used to pool the data.

after surgery, with low-certainty and very low-certainty evi- carried out in 1 study69 revealed a stronger difference in the
dence, respectively: the mean differences were −0.71 (95% CI, acupuncture group compared with controls, with a mean dif-
−1.44 to 0.02 opioids in milligrams per hour; P = .06; I2 = 64%) ference of 57.90 (95% CI, 6.52-99.28 minutes; P = .006; I2 = not
for acupuncture (eFigure 7 in the Supplement) and 6.58 (95% applicable).
CI, −6.33, to 19.49 opioids in milligrams per hour; P = .32;
I2 = 87%) for CPM (eFigure 8 in the Supplement). Preoperative Conflict of Interest of Included Studies
exercise studies did not report data on opioid consumption. Authors of 7 studies reported at least 1 conflict of interest state-
To address opioid consumption changes across the study ment. Only 3 studies32,33,49 explicitly identified the funding
period, we conducted sensitivity analyses stratifying for pe- sources (eTable 3 in the Supplement).
riod (before or after 2000). The results were not significant
(eFigure 39 and eFigure 40 in the Supplement).

Discussion
Time Before the First Analgesic Treatment
In 2 studies (124 patients), we assessed time to first PCA in the This meta-analysis found moderate evidence that electro-
acupuncture group and found moderate-certainty evidence therapy and acupuncture improved postoperative pain
that acupuncture significantly increases this period (mean dif- management and reduced opioid consumption. We found very
ference, 46.17; 95% CI, 20.84-71.50 minutes; P < .001; I2 = 19%) low-certainty evidence that cryotherapy reduced opioid
(eFigure 9 in the Supplement). Also, a subgroup analysis consumption, but there was no evidence that it improves

E10 JAMA Surgery Published online August 16, 2017 (Reprinted) jamasurgery.com

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

 Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty Original Investigation Research

perceived pain. The meta-analysis suggests that CPM and is an expensive and time-consuming procedure.33,71 Because
preoperative exercise do not help alleviate pain (measured at the results of other studies have suggested that CPM is ineffec-
different time points and using different scales) or reduce tive in improving functionality11 and that CPM is associated with
opioid consumption. increased hospital length of stay,71 careful consideration should
Electrotherapy and acupuncture are known to reduce post- be exercised before applying this treatment.
operative pain. Electrotherapy is thought to decrease pain by Our study also found little evidence to support that pre-
stimulating the pain fibers with a nonpainful stimulus that operative exercise improves postoperative pain and thus adds
blocks painful stimuli from reaching the brain and is free of to conflicting literature. Several studies have reported that pre-
adverse effects.10 One study70 recommended electrotherapy operative exercise had no significant benefit in improving func-
to reduce analgesic use for various surgical procedures. Our tionality, quality of life, or pain for patients after TKA,72,73
findings suggest that electrotherapy may not only reduce early whereas others found that the intervention improved postop-
pain but also change the long-term trajectory of recovery from erative pain, hospital length of stay, and physical function af-
pain after TKA. We found evidence that electrotherapy changed ter various surgical procedures.74 However, given the poor qual-
pain severity at 1, 2, and 6 months, with increasing effect sizes ity of the evidence, our results do not support the use of
over time. Hence, electrotherapy might be considered an ef- preoperative exercise for patients after TKA and advocate for
fective nonpharmacological ancillary intervention to stan- further high-quality studies on this topic.
dard pharmacological therapy for long-term pain improve-
ment. This finding is an important and underappreciated Limitations
contribution to the literature that examines factors influenc- Several limitations should be considered before interpreting
ing the propensity to develop chronic pain after surgery, an area these findings. First, for each intervention and outcome, we
of significant general interest in clinical literature.4 However, could only include a small number of studies in the analysis be-
because the quality of the studies analyzed for this outcome cause of high heterogeneity in the timing and type of interven-
was very low, more high-quality RCTs on long-term pain tions. To address this issue, we pooled studies from different
improvement after electrotherapy are needed. time points to obtain larger sample sizes, and subgroup analy-
Our findings showed that acupuncture pain relief ben- ses showed results similar to the overall findings (eFigures 30,
efits concentrate in the early postoperative phase but are 31, 32, 33, 34, 35, 36, 37, and 38 in the Supplement). Age and sex
ineffective in the long run. A delay in opioid consumption can were not differently distributed in the groups (treatment vs con-
be considered a proxy of lower pain levels; high postopera- trol as shown in the meta-regressions) (eFigure 41 and eFigure
tive pain can lead to chronic pain.3 Our results suggest that acu- 42 in the Supplement). Second, studies often showed a high risk
puncture led to a modest delay in PCA requests, leading to pos- or unclear risk of bias, which may have led to overestimations
sible benefits in this critical time window. Similarly, others have or underestimations of the reported effects. However, we as-
found that acupuncture provides significant pain improve- sessed the quality of evidence through a validated tool and took
ment in patients undergoing TKA and total hip arthroplasty in into account the level of certainty of evidence for each out-
the first 2 days after surgery.16 The acupuncture studies had come. Third, most studies did not achieve full masking, which
less risk of bias than other modalities, so our conclusions re- also may have caused overestimation of effects in various meta-
garding their benefit are more secure. If confirmed in future analyses conducted. Fourth, some studies lacked sufficient data
studies, our findings support the use of both electrotherapy to measure dispersion for the effect measurement (SD or SE).
and acupuncture after TKA. We attempted to address this problem by contacting authors but
We found less evidence that cryotherapy reduced opioid never obtained a response.
and NSAID consumption. While a Cochrane review article re-
ported a small benefit of cryotherapy for pain at 2 days after
surgery but not at 1 and 3 days,23 our results demonstrated very
low-certainty evidence for this intervention on postopera-
Conclusions
tive analgesia after TKA. More research about this interven- Although past studies8,10,11,16,23,73 have investigated indi-
tion could focus on opioid consumption effects. vidual nonpharmaceutical interventions for different postop-
The CPM results are particularly notable. Continuous erative outcomes after TKA, to our knowledge, this meta-
passive motion is commonly used after TKA, with the 2 analysis is the first comprehensive study to examine the most
proposed benefits of improved function and reduced pain. frequent treatments, adding new evidence on drug consump-
However, recent work has not shown the usefulness of CPM tion. As prescription opioid use is under national scrutiny and
in improving functionality and rehabilitation.11 The RCTs in- because surgery has been identified as an avenue for addic-
cluded in our meta-analysis found very low-certainty evi- tion, it is important to recognize effective alternatives to stan-
dence that CPM reduces opioid consumption during the early dard pharmacological therapy, which remains the first op-
postoperative phase and found low-certainty or very low- tion for treatment.5,12 Our study provides modest but clinically
certainty evidence that CPM provides no improvement in significant evidence that electrotherapy and acupuncture can
perceived pain. Our findings are consistent with a Cochrane potentially reduce and delay opioid consumption. However,
review article that also found no benefits of CPM on function, strong supporting research is further needed. Evidence for
pain, or quality of life after TKA.11 These results need to be cau- other interventions, although limited by the quality of the un-
tiously considered because CPM is not without risk.71 Also, CPM derlying literature, had less support.

jamasurgery.com (Reprinted) JAMA Surgery Published online August 16, 2017 E11

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

 Research Original Investigation Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty

ARTICLE INFORMATION 5. Doctor J, Cowan P, Meeker D, Bruckenthal P, /PROSPERO/display_record.asp?ID

Accepted for Publication: April 29, 2017. Broderick J. Encouraging integrative, non-opioid =CRD42016052104. Accessed November 28, 2016.
approaches to pain: a policy agenda. http: 20. Greenhalgh T, Peacock R. Effectiveness and
Published Online: August 16, 2017. //healthaffairs.org/blog/2016/10/04/encouraging-
doi:10.1001/jamasurg.2017.2872 efficiency of search methods in systematic reviews
integrative-non-opioid-approaches of complex evidence: audit of primary sources. BMJ.
Author Affiliations: Department of Medicine, -to-pain-a-policy-agenda/. Accessed November 17, 2005;331(7524):1064-1065.
Stanford University, Stanford, California (Tedesco, 2016.
Desai, Asch, McDonald, Hernandez-Boussard); 21. Sheppard MS, Westlake SM, McQuarrie A.
6. Kerr DR, Kohan L. Local infiltration analgesia: Continuous passive motion: where are we now?
Department of Biomedical and Neuromotor a technique for the control of acute postoperative
Sciences, University of Bologna, Bologna, Italy Physiother Can. 1995;47(1):36-39.
pain following knee and hip surgery: a case study of
(Tedesco, Gori, Fantini); Center for Innovation to 325 patients. Acta Orthop. 2008;79(2):174-183. 22. Beaupré LA, Lier D, Davies DM, Johnston DB.
Implementation, Veterans Affairs Palo Alto Health The effect of a preoperative exercise and education
Care System, Palo Alto, California (Asch); 7. Zukowski M, Kotfis K. The use of opioid program on functional recovery, health related
Department of Anesthesiology, Perioperative and adjuvants in perioperative multimodal analgesia. quality of life, and health service utilization
Pain Medicine, Stanford University, Stanford, Anaesthesiol Intensive Ther. 2012;44(1):42-46. following primary total knee arthroplasty.
California (Carroll); Department of Surgery, 8. Chughtai M, Elmallah RD, Mistry JB, et al. J Rheumatol. 2004;31(6):1166-1173.
Veterans Affairs Palo Alto Health Care System, Palo Nonpharmacologic pain management and muscle 23. Adie S, Kwan A, Naylor JM, Harris IA, Mittal R.
Alto, California (Curtin); Department of Surgery, strengthening following total knee arthroplasty. Cryotherapy following total knee replacement.
Stanford University, Stanford, California (Curtin, J Knee Surg. 2016;29(3):194-200. Cochrane Database Syst Rev. 2012;(9):CD007911.
Hernandez-Boussard); Department of Biomedical 9. Shakespeare D, Kinzel V. Rehabilitation after
Data Science, Stanford University, California 24. Watson T. Key concepts with electrophysical
total knee replacement: time to go home? Knee. agents. Phys Ther Rev. 2010;15(4):351-359.
(Hernandez-Boussard). 2005;12(3):185-189.
Author Contributions: Dr Hernandez-Boussard 25. Sims J. The mechanism of acupuncture
10. Mascarin NC, Vancini RL, Andrade ML, analgesia: a review. Complement Ther Med. 1997;5
had full access to all of the data in the study and Magalhães EP, de Lira CA, Coimbra IB. Effects of
takes responsibility for the integrity of the data and (2):102-111. doi:10.1016/S0965-2299(97)80008-8
kinesiotherapy, ultrasound and electrotherapy in
the accuracy of the data analysis. management of bilateral knee osteoarthritis: 26. Harris RE, Zubieta JK, Scott DJ, Napadow V,
Study concept and design: Tedesco, Asch, Carroll, prospective clinical trial. BMC Musculoskelet Disord. Gracely RH, Clauw DJ. Traditional Chinese
Curtin, McDonald, Fantini, Hernandez-Boussard. 2012;13(13):182. acupuncture and placebo (sham) acupuncture are
Acquisition, analysis, or interpretation of data: differentiated by their effects on μ-opioid receptors
Tedesco, Gori, Desai, Curtin, Fantini, 11. Harvey LA, Brosseau L, Herbert RD. Continuous (MORs). Neuroimage. 2009;47(3):1077-1085.
Hernandez-Boussard. passive motion following total knee arthroplasty in
people with arthritis. Cochrane Database Syst Rev. 27. Higgins JPT, Green S, eds. Cochrane Handbook
Drafting of the manuscript: Tedesco, Asch, Curtin. for Systematic Reviews of Interventions. Version
Critical revision of the manuscript for important 2014;(2):CD004260.
5.1.0. http://handbook.cochrane.org. Updated
intellectual content: All authors. 12. Abbasi J. As opioid epidemic rages, March 2011. Accessed July 10, 2016.
Statistical analysis: Tedesco, Gori. complementary health approaches to pain gain
Obtained funding: Hernandez-Boussard. traction. JAMA. 2016;316(22):2343-2344. 28. Guyatt GH, Oxman AD, Schünemann HJ,
Administrative, technical, or material support: Tugwell P, Knottnerus A. GRADE guidelines: a new
13. Brander V, Stulberg SD. Rehabilitation after series of articles in the Journal of Clinical
Fantini, Hernandez-Boussard. hip- and knee-joint replacement: an experience-
Study supervision: Asch, Carroll, Curtin, McDonald, Epidemiology. J Clin Epidemiol. 2011;64(4):380-382.
and evidence-based approach to care. Am J Phys
Fantini, Hernandez-Boussard. Med Rehabil. 2006;85(11)(suppl):S98-S118. 29. Furlan AD, Malmivaara A, Chou R, et al;
Conflict of Interest Disclosures: None reported. Editorial Board of the Cochrane Back, Neck Group.
14. Beswick AD, Wylde V, Gooberman-Hill R, Blom 2015 Updated Method Guideline for Systematic
Funding/Support: This project was supported by A, Dieppe P. What proportion of patients report Reviews in the Cochrane Back and Neck Group.
grant R01HS024096 from the Agency for long-term pain after total hip or knee replacement Spine (Phila Pa 1976). 2015;40(21):1660-1673.
Healthcare Research and Quality (to principal for osteoarthritis? a systematic review of
investigator Dr Hernandez-Boussard). prospective studies in unselected patients. BMJ Open. 30. DerSimonian R, Laird N. Meta-analysis in
2012;2(1):e000435-e000435. clinical trials. Control Clin Trials. 1986;7(3):177-188.
Role of the Funder/Sponsor: The funding source
had no role in the design and conduct of the study; 15. Puolakka PA, Rorarius MG, Roviola M, Puolakka 31. Beaupré LA, Davies DM, Jones CA, Cinats JG.
collection, management, analysis, and TJ, Nordhausen K, Lindgren L. Persistent pain Exercise combined with continuous passive motion
interpretation of the data; preparation, review, or following knee arthroplasty. Eur J Anaesthesiol. or slider board therapy compared with exercise
approval of the manuscript; and decision to submit 2010;27(5):455-460. only: a randomized controlled trial of patients
the manuscript for publication. following total knee arthroplasty. Phys Ther. 2001;
16. Crespin DJ, Griffin KH, Johnson JR, et al. 81(4):1029-1037.
Acupuncture provides short-term pain relief for
REFERENCES patients in a total joint replacement program. Pain 32. Bennett LA, Brearley SC, Hart JA, Bailey MJ.
1. Weiser TG, Regenbogen SE, Thompson KD, et al. Med. 2015;16(6):1195-1203. A comparison of 2 continuous passive motion
An estimation of the global volume of surgery: protocols after total knee arthroplasty: a controlled
17. McCartney CJ, Nelligan K. Postoperative pain and randomized study. J Arthroplasty. 2005;20
a modelling strategy based on available data. Lancet. management after total knee arthroplasty in elderly
2008;372(9633):139-144. (2):225-233.
patients: treatment options. Drugs Aging. 2014;31
2. Apfelbaum JL, Chen C, Mehta SS, Gan TJ. (2):83-91. 33. Bruun-Olsen V, Heiberg KE, Mengshoel AM.
Postoperative pain experience: results from a Continuous passive motion as an adjunct to active
18. Liberati A, Altman DG, Tetzlaff J, et al. The exercises in early rehabilitation following total knee
national survey suggest postoperative pain PRISMA statement for reporting systematic reviews
continues to be undermanaged. Anesth Analg. arthroplasty: a randomized controlled trial. Disabil
and meta-analyses of studies that evaluate health Rehabil. 2009;31(4):277-283.
2003;97(2):534-540. care interventions: explanation and elaboration.
3. Joshi GP, Ogunnaike BO. Consequences of J Clin Epidemiol. 2009;62(10):e1-e34. 34. Chen LH, Chen CH, Lin SY, et al. Aggressive
inadequate postoperative pain relief and chronic continuous passive motion exercise does not
19. Gori D, Tedesco D, Fantini MP, improve knee range of motion after total knee
persistent postoperative pain. Anesthesiol Clin Hernandez-Boussard T, Desai K, Bianciardi L.
North America. 2005;23(1):21-36. arthroplasty. J Clin Nurs. 2013;22(3-4):389-394.
Non-pharmaceutical interventions in
4. Kehlet H, Jensen TS, Woolf CJ. Persistent post-operative pain management of total knee 35. Colwell CW Jr, Morris BA. The influence of
postsurgical pain: risk factors and prevention. Lancet. replacement. http://www.crd.york.ac.uk continuous passive motion on the results of total
2006;367(9522):1618-1625.

E12 JAMA Surgery Published online August 16, 2017 (Reprinted) jamasurgery.com

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017

 Drug-Free Interventions to Reduce Pain or Opioid Consumption After Total Knee Arthroplasty Original Investigation Research

knee arthroplasty. Clin Orthop Relat Res. 1992; 49. Calatayud J, Casaña J, Ezzatvar Y, Jakobsen 63. Adravanti P, Nicoletti S, Setti S, Ampollini A,
(276):225-228. MD, Sundstrup E, Andersen LL. High-intensity de Girolamo L. Effect of pulsed electromagnetic
36. Denis M, Moffet H, Caron F, Ouellet D, Paquet preoperative training improves physical and field therapy in patients undergoing total knee
J, Nolet L. Effectiveness of continuous passive functional recovery in the early post-operative arthroplasty: a randomised controlled trial. Int Orthop.
motion and conventional physical therapy after periods after total knee arthroplasty: a randomized 2014;38(2):397-403.
total knee arthroplasty: a randomized clinical trial. controlled trial [published online January 14, 2016]. 64. Borckardt JJ, Reeves ST, Robinson SM, et al.
Phys Ther. 2006;86(2):174-185. Knee Surg Sports Traumatol Arthrosc. Transcranial direct current stimulation (tDCS)
37. Harms M, Engstrom B. Continuous passive 50. Gstoettner M, Raschner C, Dirnberger E, reduces postsurgical opioid consumption in total
motion as an adjunct to treatment in the Leimser H, Krismer M. Preoperative proprioceptive knee arthroplasty (TKA). Clin J Pain. 2013;29(11):
physiotherapy management of the total knee training in patients with total knee arthroplasty. Knee. 925-928.
arthroplasty patient. Physiotherapy. 1991;77(4): 2011;18(4):265-270. 65. Moretti B, Notarnicola A, Moretti L, et al. I-ONE
301-307. doi:10.1016/S0031-9406(10)61768-3 51. McKay C, Prapavessis H, Doherty T. The effect therapy in patients undergoing total knee
38. Kim TK, Park KK, Yoon SW, Kim SJ, Chang CB, of a prehabilitation exercise program on quadriceps arthroplasty: a prospective, randomized and
Seong SC. Clinical value of regular passive ROM strength for patients undergoing total knee controlled study. BMC Musculoskelet Disord. 2012;13
exercise by a physical therapist after total knee arthroplasty: a randomized controlled pilot study. (1):88.
arthroplasty. Knee Surg Sports Traumatol Arthrosc. PM R. 2012;4(9):647-656. 66. Chen CC, Yang CC, Hu CC, Shih HN, Chang YH,
2009;17(10):1152-1158. 52. Albrecht S, le Blond R, Köhler V, et al. Hsieh PH. Acupuncture for pain relief after total
39. Lenssen A, de Bie RA, Bulstra SK, van Steyn [Cryotherapy as analgesic technique in direct, knee arthroplasty: a randomized controlled trial.
MJA. Continuous passive motion (CPM) in postoperative treatment following elective joint Reg Anesth Pain Med. 2015;40(1):31-36.
rehabilitation following total knee arthroplasty: replacement] [in German]. Z Orthop Ihre Grenzgeb. 67. Mikashima Y, Takagi T, Tomatsu T, Horikoshi M,
a randomised controlled trial. Phys Ther Rev. 2003; 1997;135(1):45-51. Ikari K, Momohara S. Efficacy of acupuncture during
(8):123-129. doi:10.1179/108331903225003019 53. Gibbons CE, Solan MC, Ricketts DM, Patterson post-acute phase of rehabilitation after total knee
40. Lenssen TA, van Steyn MJ, Crijns YH, et al. M. Cryotherapy compared with Robert Jones arthroplasty. J Tradit Chin Med. 2012;32(4):545-548.
Effectiveness of prolonged use of continuous bandage after total knee replacement: 68. Tsang RC, Tsang PL, Ko CY, Kong BC, Lee WY,
passive motion (CPM), as an adjunct to a prospective randomized trial. Int Orthop. 2001;25 Yip HT. Effects of acupuncture and sham
physiotherapy, after total knee arthroplasty. BMC (4):250-252. acupuncture in addition to physiotherapy in
Musculoskelet Disord. 2008;9:60. 54. Ivey M, Johnston RV, Uchida T. Cryotherapy for patients undergoing bilateral total knee
41. MacDonald SJ, Bourne RB, Rorabeck CH, postoperative pain relief following knee arthroplasty: a randomized controlled trial. Clin
McCalden RW, Kramer J, Vaz M. Prospective arthroplasty. J Arthroplasty. 1994;9(3):285-290. Rehabil. 2007;21(8):719-728.
randomized clinical trial of continuous passive 55. Kullenberg B, Ylipää S, Söderlund K, Resch S. 69. Tzeng CY, Chang SL, Wu CC, et al.
motion after total knee arthroplasty. Clin Orthop Postoperative cryotherapy after total knee Single-blinded, randomised preliminary study
Relat Res. 2000;(380):30-35. arthroplasty: a prospective study of 86 patients. evaluating the effects of 2 Hz electroacupuncture
42. Maniar RN, Baviskar JV, Singhi T, Rathi SS. To J Arthroplasty. 2006;21(8):1175-1179. for postoperative pain in patients with total knee
use or not to use continuous passive motion 56. Levy AS, Marmar E. The role of cold arthroplasty. Acupunct Med. 2015;33(4):284-288.
post–total knee arthroplasty presenting functional compression dressings in the postoperative 70. Bjordal JM, Johnson MI, Ljunggreen AE.
assessment results in early recovery. J Arthroplasty. treatment of total knee arthroplasty. Clin Orthop Transcutaneous electrical nerve stimulation (TENS)
2012;27(2):193-200.e1. Relat Res. 1993;(297):174-178. can reduce postoperative analgesic consumption:
43. May LA, Busse W, Zayac D, Whitridge MR. 57. Morsi E. Continuous-flow cold therapy after a meta-analysis with assessment of optimal
Comparison of continuous passive motion (CPM) total knee arthroplasty. J Arthroplasty. 2002;17(6): treatment parameters for postoperative pain. Eur J
machines and lower limb mobility boards (LLiMB) in 718-722. Pain. 2003;7(2):181-188.
the rehabilitation of patients with total knee 58. Radkowski CA, Pietrobon R, Vail TP, Nunley JA 71. Joshi RN, White PB, Murray-Weir M, Alexiades
arthroplasty. Can J Rehabil. 1999;12:257-263. II, Jain NB, Easley ME. Cryotherapy temperature MM, Sculco TP, Ranawat AS. Prospective
44. McInnes J, Larson MG, Daltroy LH, et al. differences after total knee arthroplasty: randomized trial of the efficacy of continuous
A controlled evaluation of continuous passive a prospective randomized trial. J Surg Orthop Adv. passive motion post total knee arthroplasty:
motion in patients undergoing total knee 2007;16(2):67-72. experience of the Hospital for Special Surgery.
arthroplasty. JAMA. 1992;268(11):1423-1428. J Arthroplasty. 2015;30(12):2364-2369.
59. Smith J, Stevens J, Taylor M, Tibbey J.
45. Montgomery F, Eliasson M. Continuous passive Bandaging and cold therapy in total knee 72. Cabilan CJ, Hines S, Munday J. The
motion compared to active physical therapy after replacement surgery. Orthop Nurs. 2002;21(2):61- effectiveness of prehabilitation or preoperative
knee arthroplasty: similar hospitalization times in a 66. exercise for surgical patients: a systematic review.
randomized study of 68 patients. Acta Orthop Scand. JBI Database System Rev Implement Rep. 2015;13(1):
60. Su EP, Perna M, Boettner F, et al. 146-187.
1996;67(1):7-9. A prospective, multi-center, randomised trial to
46. Pope RO, Corcoran S, McCaul K, Howie DW. evaluate the efficacy of a cryopneumatic device on 73. Hoogeboom TJ, Oosting E, Vriezekolk JE, et al.
Continuous passive motion after primary total knee total knee arthroplasty recovery. J Bone Joint Surg Br. Therapeutic validity and effectiveness of
arthroplasty: does it offer any benefits? J Bone Joint 2012;94(11)(suppl A):153-156. preoperative exercise on functional recovery after
Surg Br. 1997;79(6):914-917. joint replacement: a systematic review and
61. Thienpont E. Does advanced cryotherapy meta-analysis. PLoS One. 2012;7(5):e38031.
47. Sahin E, Akalin E, Bircan C, et al. The effects of reduce pain and narcotic consumption after knee
continuous passive motion on outcome in total arthroplasty? Clin Orthop Relat Res. 2014;472(11): 74. Santa Mina D, Clarke H, Ritvo P, et al. Effect of
knee arthroplasty. J Rheumatol Med Rehabil. 2006; 3417-3423. total-body prehabilitation on postoperative
17(2):85-90. outcomes: a systematic review and meta-analysis.
62. Webb JM, Williams D, Ivory JP, Day S, Physiotherapy. 2014;100(3):196-207.
48. Walker RH, Morris BA, Angulo DL, Schneider J, Williamson DM. The use of cold compression
Colwell CW Jr. Postoperative use of continuous dressings after total knee replacement:
passive motion, transcutaneous electrical nerve a randomized controlled trial. Orthopedics. 1998;21
stimulation, and continuous cooling pad following (1):59-61.
total knee arthroplasty. J Arthroplasty. 1991;6(2):
151-156.

jamasurgery.com (Reprinted) JAMA Surgery Published online August 16, 2017 E13

© 2017 American Medical Association. All rights reserved.

Downloaded From: http://archsurg.jamanetwork.com/ by a Columbia University User on 08/17/2017