Original Research

Accuracy of Ultrasound-Guided and Palpation-Guided Knee Injections By an Experienced and Less-Experienced Injector Using a Superolateral Approach: A Cadaveric Study
Heather M. Curtiss, MD, Jonathan T. Finnoff, DO, Evan Peck, MD, John Hollman, PT, PhD, Jeff Muir, MD, Jay Smith, MD
Objectives: To evaluate the accuracy of ultrasound (US)-guided and palpation-guided knee injections by an experienced and a less-experienced clinician with use of a superolateral approach. Design: Single-blinded, prospective study. Setting: Academic institution procedural skills laboratory. Participants: Twenty cadaveric knee specimens without trauma, surgery, or major deformity. Intervention: US-guided and palpation-guided knee injections of colored liquid latex were performed in each specimen by an experienced and a less-experienced clinician with use of a superolateral approach. The order of injections was randomized. The specimens were subsequently dissected by a blinded investigator and assessed for accuracy. Main Outcomes: Accuracy was divided into 3 categories: (1) accurate (all of the injectate was within the joint), (2) partially accurate (some of the injectate was within the joint and some was within the periarticular tissues), and (3) inaccurate (none of the injectate was within the joint). The accuracy rates were calculated for each clinician and guidance method. Results: US-guided knee injections that used a superolateral approach were 100% accurate for both clinicians. Palpation-guided knee injections that used a superolateral approach were signicantly inuenced by experience, with the less-experienced investigator demonstrating an accuracy rate of 55% (95% condence interval 34%-74%) and the more experienced investigator demonstrating an accuracy rate of 100% (95% condence interval 81%-100%). Conclusions: US-guided knee injections that use a superolateral approach are very accurate in a cadaveric model, whereas the accuracy of palpation-guided knee injections that use the same approach is variable and appears to be signicantly inuenced by clinician experience. These ndings suggest that US guidance should be considered when one performs knee injections with a superolateral approach that require a high degree of accuracy. PM R 2011;3:507-515

H.M.C. Department of Physical Medicine & Rehabilitation, Mayo Clinic College of Medicine, Mayo Clinic Sports Medicine Center, Rochester, MN Disclosure: nothing to disclose J.T.F. Department of Physical Medicine & Rehabilitation, Mayo Clinic College of Medicine, Mayo Clinic Sports Medicine Center, E14 Mayo Building, 200 1st St, SW, Rochester, MN 55905. Address correspondence to J.T.F.; e-mail: Finnoff.jonathan@mayo.edu Disclosure: nothing to disclose E.P. Section of Sports Medicine, Department of Orthopaedic Surgery, Cleveland Clinic Florida, West Palm Beach, FL Disclosure: nothing to disclose J.H. Department of Physical Medicine & Rehabilitation, Mayo Clinic College of Medicine, Mayo Clinic Sports Medicine Center, Rochester, MN Disclosure: nothing to disclose J.M. Department of Physical Medicine & Rehabilitation, Mayo Clinic College of Medicine, Mayo Clinic Sports Medicine Center, Rochester, MN Disclosure: nothing to disclose J.S. Department of Physical Medicine & Rehabilitation, Mayo Clinic College of Medicine, Mayo Clinic Sports Medicine Center, Rochester, MN Disclosure: 2A, Tenex; 3A, Gulf Coast Ultrasound, Andrews Institute Submitted for publication November 18, 2010; accepted February 10, 2011.

INTRODUCTION
Knee pain is extremely common in the general population and accounts for one third of musculoskeletal complaints in the primary care setting [1]. Although the treatment options for knee pain are as diverse as the disorders that affect the knee, one frequently used therapeutic option is an intra-articular injection. Intra-articular knee injections can be used diagnostically to identify a patients source of pain or therapeutically to deliver medications (eg, corticosteroids or viscosupplements) or biologic agents (eg, platelet-rich plasma [PRP] or stem cells) to reduce pain and improve function in patients with knee disorders [2-5]. Although a great degree of accuracy is required for diagnostic injections to provide adequate sensitivity and specicity for this type of injection, a growing body of literature supports the
PM&R
1934-1482/11/$36.00 Printed in U.S.A.

2011 by the American Academy of Physical Medicine and Rehabilitation

Vol. 3, 507-515, June 2011 DOI: 10.1016/j.pmrj.2011.02.020

507

508

Curtiss et al

SONOGRAPHIC- VS. PALPATION-GUIDED KNEE INJECTION

notion that therapeutic outcome is also dependent upon injection accuracy [6-9]. This notion is particularly true for biologic agents or viscosupplement injections in which the therapeutic benet from the injection will only occur if the injectate is placed accurately within the joint [3-5]. Therefore it is essential to assess and optimize injection accuracy to maximize efcacy. Currently, research suggests that the accuracy of palpation-guided knee injections is highly variable (50%-93%) depending on the technique used [6,10-14]. Many techniques have been used in an attempt to improve the accuracy of palpation-guided knee injections, including air arthrography, triangulation techniques, and traction, but results continue to be mixed, with accuracy rates ranging from 71%91% [12,14,15]. To date, no investigators have reported a 100% accuracy rate with any palpation-guided knee injection technique. Imaging modalities, including uoroscopy and ultrasound (US), have been shown to increase the accuracy of joint injections [16-20]. Two studies reported 100% accuracy of uoroscopically guided knee injections [16,17]. However, many limitations are associated with uoroscopically guided injections, including radiation exposure to both the patient and the clinician, use of contrast, equipment expense, lack of portability, and specialized room requirements [16-22]. A portable, low-cost, and safe alternative to uoroscopy for many types of injections is US [18,21,22]. Two studies have investigated the accuracy of US-guided knee injections. In 2001, Qvistgaard et al [19] evaluated the accuracy of US-guided knee injections by using a midpatellar lateral approach in the absence of a knee effusion. Although the patella prevented direct visualization of the needle during the injection with this technique, the authors concluded that the US-guided injections were 100% accurate. Unfortunately, no criteria beyond US visualization of the injectate within the joint were used to conrm injection accuracy, and it is possible that some unrecognized periarticular injections occurred in this study. More recently, Im et al [20] compared palpation-guided and US-guided knee injections with use of a midpatellar medial approach. The injectate included 2 mL of radioopaque dye, and a postinjection radiograph was used to conrm intra-articular placement of the injectate. The authors reported a 77% accuracy rate with the palpationguided technique and a 96% accuracy rate for the US-guided technique. Although the US-guided approach increased the accuracy of the injection, once again, the technique did not allow direct visualization of the needle tip while the injectate was delivered because of obstruction of the needle tip image by the medial border of the patella. Furthermore, because the palpation-guided midpatellar medial approach has demonstrated poor accuracy in previous studies and therefore is not the technique of choice in clinical practice, it is apparent that

further research is required to compare a more optimal palpation-guided knee injection technique with a US-guided knee injection technique that addresses the limitations of the previously described techniques. To our knowledge, no studies have compared the accuracy of US-guided and palpation-guided knee injections with the use of the superolateral approach, which is the palpationguided and US-guided technique most frequently used in our practice. During US-guided knee injections, the superolateral approach allows for direct imaging of the needle, including the needle tip, throughout the entire procedure. Thus it addresses the major limitation of the 2 US-guided knee injection techniques previously described. Furthermore, in the single study in which the accuracy of palpation-guided knee injections via a superolateral approach was investigated, the authors reported the accuracy of this technique to be 90% [13]. However, because of the limited research related to this technique and previous research suggesting a high degree of variability in knee injection accuracy between studies that use the same technique, further research investigating the accuracy of palpation-guided knee injections with use of the superolateral approach is warranted. Therefore the primary purposes of this study were to evaluate the accuracy of US-guided and palpation-guided knee injections with use of a superolateral approach. In addition, to our knowledge, no studies have been performed in which the authors evaluated the effect of operator experience on knee injection accuracy. Therefore a secondary aim of this study was to evaluate the accuracy of an experienced clinician and a less experienced clinician when they used a superolateral approach for both palpation-guided and USguided knee injections.

METHODS Anatomic Specimens

Twenty unembalmed fresh-frozen adult lower limb cadaveric specimens were obtained through the Mayo Clinic Department of Anatomy Mayo Foundation Bequest Program. Demographic data from the donors can be found in Table 1. Specimens were thawed at room temperature immediately before the study. No specimen demonstrated signs of previous surgery, trauma, or major deformity. The project was approved by the Mayo Clinic Bio-Specimens Subcommittee of the Institutional Review Board.
Table 1. Cadaveric specimen demographic characteristics (n 20) Male/ Female 14/6 Age, y 75.7 (14.5) Height, cm 170.1 (10.7) Weight, kg 70.0 (13.1) BMI, kg/m2 24.2 (4.3)

Values are mean (standard deviation). BMI body mass index.

PM&R

Vol. 3, Iss. 6, 2011

509

Figure 1. Palpation-guided intra-articular knee injection at the intersection of the superior one third of the patella and lateral patellofemoral joint.

Equipment
US-guided procedures were performed with either a Philips iU22 or a Philips CX50 US machine (Philips Ultrasound Systems, Bothell, WA). The sonographic scans were performed with either a 12-5 MHz linear array (iU22) or a 12-3 MHz linear array (CX50) transducer and standard US gel. A 22-gauge, 2-inch needle was used for the injections.

use of US guidance so that the knees of all 20 cadaveric specimens were free from effusion before the knee injections were performed. Each investigator performed one palpation-guided and one US-guided knee injection into each of the 20 cadaveric specimens. The order of injection (palpation-guided versus US-guided) was randomized. For the palpation-guided technique, the lateral knee was marked with an indelible marker by 1 of the 2 injectors (either J.T.F. or E.R.P.) at the junction of the superior and middle thirds of the patella and the lateral patellofemoral joint (Figure 1). Only one mark was placed on each cadaver. Thus the needle entry point into the cadaver was the same for each injection regardless of injector. The needle was advanced toward the femoral notch in a 45 cephalo-lateral to caudo-medial direction between the femoral condyle and the lateral border of the patella. The needle was advanced until a loss of resistance was appreciated, at which time 1 mL of injectate was delivered into the knee joint. For the US-guided technique, the transducer was placed in a transverse plane over the distal aspect of the quadriceps tendon just proximal to its insertion onto the cephalad border of the patella (Figure 2). At this level, the skin, subcutaneous tissue, quadriceps tendon, quadriceps fat pad, prefemoral fat pad, and femur could be imaged. The suprapa-

Injection Procedure
Two investigators performed the injections (E.P., J.T.F.). One investigator (E.P.) was in fellowship, with approximately 3 years and 10 months of experience performing palpationguided knee injections and 10 months of experience performing US-guided knee injections at the time of data collection. The second investigator (J.T.F.) was a staff physician with approximately 13 years of experience performing palpation-guided knee injections and 3 years of experience performing US-guided knee injections at the time of data collection. The cadaveric specimens were positioned such that the patella was facing toward the ceiling and the knee was exed approximately 20. The specimen was stabilized in this position by an investigator. All of the cadaveric knees were then scanned with the US equipment described previously to determine whether a knee effusion was present. This procedure was performed by an investigator who did not perform any of the injections (H.C.). If a knee effusion was identied, it was aspirated with an 18-gauge, 1.5-inch needle with the

Figure 2. Ultrasound-guided intra-articular knee injection. The needle is visualized as it enters the suprapatellar pouch in long axis. (Note: The X on the skin is the location of the palpationguided injection.)

510

Curtiss et al

SONOGRAPHIC- VS. PALPATION-GUIDED KNEE INJECTION

blinded investigator using the following grading scale: (1) accurate (all of the injectate within the joint); (2) partially accurate (some of the injectate in the joint and some periarticular); and (3) inaccurate (none of the injectate within the joint; Figure 4).

Statistics
For statistical analysis purposes, the data were classied in binomial groups, separated in 2 different ways. The rst analysis considered the accurate group individually, whereas the partially accurate and inaccurate groups were combined (analysis 1). The second analysis considered the inaccurate group individually, whereas the accurate and partially accu-

Figure 3. Ultrasound picture demonstrating the needle within the suprapatellar recess during an ultrasound-guided injection. The quadriceps tendon is above the needle, and the pre-femoral fat pad is below the needle. N needle; PF prefemoral fat pad.

tellar recess lies between the quadriceps fat pad and the prefemoral fat pad at this level. The transducer was glided proximally until the quadriceps fat pad was no longer present (approximately 2-4 cm), at which time the suprapatellar recess was imaged between the deep surface of the quadriceps tendon and the supercial surface of the prefemoral fat pad. Identication of the suprapatellar recess was assisted through medial and lateral mobilizations of the patella and/or prefemoral fat pad. After identifying the suprapatellar recess, the needle was introduced in a lateral-to-medial direction, in a long-axis relative to the transducer, and advanced into the suprapatellar recess (Figure 3). After imaging the needle tip in the suprapatellar recess, 1 mL of injectate was delivered into the suprapatellar recess. The injectate was a 50% diluted, colored latex solution. A different color of liquid latex was used for each investigator, and injection technique (palpation-guided versus US-guided); thus 4 different colors of latex were used. One milliliter of injectate was delivered with each injection; as a result, the entire volume of injectate within the knee joint after completion of the 4 injections was 4 mL. This volume of injectate was chosen because previous authors [23] demonstrated an inability to image knee effusions with less than 4 mL of uid, and a pilot study performed by the investigators conrmed that 4 mL of uid within the knee joint could not be detected with US. This nding prevented the study from being confounded by previous injections into the knee.

Postprocedure Evaluation
Approximately 1 week after the injections, the cadaveric specimens were dissected and graded for accuracy by a

Figure 4. Cadaveric knee dissection demonstrating latex within the joint. The patellar tendon has been cut, and the patella has been reected proximally. L latex dye; MFC medial femoral condyle; P patella.

PM&R

Vol. 3, Iss. 6, 2011

511

Table 2. Analysis 1: accuracy rates of ultrasound-guided and palpation-guided injections Injection Condition Guided Unguided *P .004. Accuracy Rate (%) 100 78* 95% Condence Interval (%) 89-100 62-88

Table 4. Analysis 1: accuracy rates of ultrasound-guided and palpation-guided injections for experienced and less-experienced clinicians Injection Administrator Ultrasound guided Experienced Less experienced Unguided Experienced Less experienced *P .004. Accuracy Rate (%) 100 100 100 55* 95% Condence Interval (%) 81-100 81-100 81-100 34-74

rate groups were combined (analysis 2). The reason for the 2 separate analyses was to provide the following clinically relevant information. If a physician is performing an injection for diagnostic purposes or is injecting a substance that requires intra-articular placement for therapeutic efcacy (eg, a biologic agent or viscosupplements), then delivery of the entire injectate within the joint is required, and thus accuracy is most appropriately dened by analysis 1. However, if a physician is not performing a diagnostic injection or is injecting a therapeutic substance that may provide benet even if some of the medication is periarticular (eg, corticosteroids), analysis 2 is a better descriptor of accurate. The accuracy rates of each injection technique (US-guided and palpation-guided) for each investigator (experienced and less experienced) were reported separately. Injection accuracy rates were reported with 95% binomial condence intervals (95% CI) calculated by means of the modied Wald method [24]. US-guided and palpation-guided injection accuracy rates and experienced and less-experienced investigator accuracy rates were calculated with McNemar tests ( 0.05) with the use of SPSS 15.0 for Windows software (SPSS Inc, Chicago, IL).

RESULTS
The results for analysis 1 are summarized in Tables 2-4. US-guided knee injections were 100% accurate (95% CI 89%-100%), whereas palpation-guided knee injections were 78% accurate (95% CI 62%-88%). The difference in accuracy between the 2 guidance techniques was statistically signicant (P .004). As shown in Table 3, experience also inuenced accuracy. The less-experienced injector had a combined palpation-guided and US-guided knee injection accuracy rate of 78% (95% CI 62%-88%) compared with a 100% accuracy rate (95% CI 89%-100%) by the more experienced investigator (P .004). However, this nding can be attributed to the difference in the palpation-guided
Table 3. Analysis 1: accuracy rates of injections for an experienced and less-experienced clinician Injection Administrator Experienced Less experienced *P .004. Accuracy Rate (%) 100 78* 95% Condence Interval (%) 89-100 62-88

accuracy rates between the more experienced (100% accurate, 95% CI 81%-100%) and less experienced (55% accurate, 95% CI 34%-74%) investigators (P .004), because the US-guided knee injection accuracy rates for both investigators was 100% (95% CI 81%-100%). Therefore experience affected the accuracy rate of the palpation-guided knee injections but not the US-guided knee injections. The results for analysis 2 are summarized in Tables 5-7. None of the results from analysis 2 demonstrated statistical signicance. However, trends similar to analysis 1 were present. The US-guided injection accuracy rate was 100% (95% CI 89%-100%), whereas palpation-guided injections were 88% accurate (95% CI 73%-95%, P .063). Therefore 12% of palpation-guided knee injections did not place any injectate into the intra-articular space, which may have clinical signicance especially for injectates that require intraarticular placement to provide therapeutic benet. The experienced injector demonstrated an overall accuracy rate of 100% (95% CI 89%-100%), whereas the less experienced injector demonstrated 88% accuracy (95% CI 73%-95%, P .063). As noted in analysis 1, because both injectors were 100% accurate (95% CI 81%-100%) when performing US-guided knee injections, the between injector accuracy difference was based entirely upon the palpation-guided knee injection accuracy rates, which were 100% (95% CI 81%-100%) for the experienced investigator versus 75% (95% CI 53%-89%) for the less experienced investigator (P .063).

DISCUSSION
To our knowledge, we are the rst to assess, with the use of a superolateral approach, the accuracy of US-guided knee inTable 5. Analysis 2: accuracy rates of ultrasound-guided and palpation-guided injections Injection Condition Guided Unguided *P .004. Accuracy Rate (%) 100 88* 95% Condence Interval (%) 89-100 73-95

512

Curtiss et al

SONOGRAPHIC- VS. PALPATION-GUIDED KNEE INJECTION

Table 6. Analysis 2: accuracy rates of injections for an experienced and less-experienced clinician Injection Administrator Experienced Less experienced *P .004. Accuracy Rate (%) 100 88* 95% Condence Interval (%) 89-100 73-95

jections into a knee without an effusion to compare the accuracy rates between palpation-guided and US-guided knee injections and to determine whether the investigators experience with injections affected their accuracy. The ndings of our study demonstrated 100% accuracy for the USguided knee injections when the superolateral approach was used, which was not affected by the duration of clinical experience. However, the accuracy of palpation-guided knee injections when a superolateral approach was used was signicantly inuenced by the investigators experience, with the less-experienced investigator demonstrating an accuracy rate of 55%-75%, depending upon the denition of accuracy, and the more experienced investigator demonstrating an accuracy rate of 100%. Therefore when a high degree of accuracy is required, our ndings suggest that less-experienced clinicians should consider using US guidance to improve the accuracy of their knee injections. Previous investigators [6,10-14] have reported accuracy rates of 50%-93% for palpation-guided knee injections. The combined accuracy rate of the experienced and less-experienced investigators for palpation-guided knee injections in this study was 78%, which is in line with the accuracy rates of previous research. However, accuracy rates for palpationguided knee injections vary depending upon the injection technique and the investigator performing the injection. In a study by Jackson et al [10], the midpatellar lateral approach demonstrated the greatest level of accuracy, which was 93%. However, the accuracy for the midpatellar lateral approach reported in the studies by Esenyel et al [11] and Wind et al [13] were 76% and 50%, respectively, thus demonstrating a high degree of variability when this injection technique is used by different investigators. In contrast, the anteromedial and anterolateral approaches demonstrate more consistent accuracy rates (71%-86%), but their accuracy rates are suboptimal [10-12]. The only study that has investigated the accuracy of palpation-guided knee injections with use of the superomedial approach reported an accuracy rate of 90% [13]. The same investigator reported a similar accuracy rate for palpationguided injections with use of the superolateral approach. Although these results were promising, the injections were performed on anesthetized patients with an 18-gauge needle, which does not represent the typical clinical setting for knee injections. This situation may have led to an overestimate in their knee injection accuracy.

Another study reported an accuracy rate of 85%-95% for palpation-guided knee injections with a superolateral approach, but they determined accuracy by a postprocedure questionnaire of clinical efcacy and did not rely upon more objective accuracy determination methods, which calls their results into question [14]. The ndings of our study conrm that the accuracy of palpation-guided knee injections is highly variable and suggest that one signicant contributing factor is the injectors experience. The accuracy rate for the less-experienced investigator when performing palpation-guided knee injections using a superolateral approach was 55%-75%, depending on how accuracy was dened, whereas the accuracy rate for the more experienced investigator was 100%. Therefore when accuracy is imperative, less experienced clinicians should avoid performing palpation-guided knee injections, and more experienced clinicians should consider using the superolateral approach when a performing a palpationguided knee injection. The accuracy rate in this study for US-guided knee injections with the use of a superolateral approach was 100%, regardless of the injectors experience. These ndings conrm those of previous studies in which the authors indicated that US-guided knee injections are extremely accurate [19,20]. However, in contrast to the US-guided knee injection techniques previously described, the technique used in this study allows direct visualization of the needle tip throughout the entire procedure, including when the injectate is delivered. Therefore, unless the patient has a contraindication to this technique, we suggest use of this technique when performing US-guided knee injections. Currently, few data are available to address the role of experience in US-guided joint injections. Intuitively, one would suspect that there is a learning curve. This learning curve has been demonstrated in studies in which authors evaluated the use of US in other diagnostic and procedural situations. For example, a recent study suggested that a minimum of 200 procedures was necessary to become procient at ne-needle aspiration of thyroid nodules based on the quality of samples obtained [25]. Van Holsbeke et al [26], who investigated the role of experience in a sonographers ability to assess adnexal masses, demonstrated that accuracy
Table 7. Analysis 2: accuracy rates of ultrasound-guided and palpation-guided injections for experienced and less experienced injectors Injection Administrator Ultrasound guided Experienced Less experienced Unguided Experienced Less experienced *P .063. Accuracy Rate (%) 100 100 100 75* 95% Condence Interval (%) 81-100 81-100 81-100 53-89

PM&R

Vol. 3, Iss. 6, 2011

513

and diagnostic condence increased with experience. It is important to note that in our study, the less-experienced investigator had nearly 1 year of experience performing approximately 10 US-guided procedures per week. Therefore the less-experienced investigators lack of experience needs to be considered in context, because the acquisition of this level of experience may take clinicians longer or shorter durations based on the characteristics of their practice. The exact duration of time or number of procedures required to achieve 100% accuracy for US-guided knee injections with use of a superolateral approach was not determined in this study and warrants future research. Injection accuracy is clinically relevant for several reasons. Although injection accuracy is not specic to the knee, several investigators have demonstrated that accurately placed injections may be more efcacious than inaccurately placed injections, have lower complications, result in less procedural and postprocedural pain, and enable a larger volume of uid to be aspirated during arthrocentesis [6-9]. In addition, the therapeutic benet of certain medications is predicated upon their accurate placement within the joint. For instance, the proposed mechanism of action for viscosupplementation is enhanced synovial uid viscosity and stimulation of increased synovial uid production, both of which require intra-articular placement of the medication [3]. Therefore an inaccurately placed viscosupplement injection for knee osteoarthritis may result in a lack of therapeutic efcacy. A study by Waddell et al [27] demonstrated a 2.1-year mean delay in knee replacement surgery through the use of intraarticular viscosupplement injections. Theoretically, because the authors did not use any form of guidance to conrm intra-articular placement of the injectate within the knee, it is possible that by increasing the injection accuracy through the use of guided injections, the delay in surgery would be even greater. Similarly, research is being conducted on the use of many promising new biologic treatments for joint injuries such as the intra-articular injection of isolated growth factors, autologous conditioned serum, preparation rich in growth factors, or PRP into injured or arthritic joints. In preliminary studies in arthritic human knees, investigators have demonstrated functional improvements and reduced knee pain after the injection of PRP, whereas in studies in arthritic rabbit knees, investigators suggest that PRP may slow the progression of knee osteoarthritis [4,5]. However, these benets are predicated upon the accurate placement of the injectate within the joint. Another benet of injection accuracy is the minimization of some procedural risks and adverse effects. It has been suggested that periarticular placement of viscosupplements may contribute to postinjection pain [6,28]. US also enables the visualization of periarticular soft-tissue structures, which allows the clinician to avoid piercing tissues and structures that should be avoided (ie, neurovascular structures). Fur-

thermore, because tissues and uid can be imaged with US, US can be used to detect synovitis, guide synovial biopsies, and guide the aspiration of joint uid for diagnostic and therapeutic purposes. For these reasons, US-guided knee injections may prove clinically usefully for reducing procedure-related complications and adverse effects and for improving diagnostic capabilities and enhancing therapeutic efcacy. This study had several limitations that warrant discussion. The rst consideration is the sample size. On the basis of a post-hoc power analysis, a sample size of 30-80 specimens (depending upon the denition of accuracy) would have been required to achieve 80% power. We were only able to obtain 20 specimens at the time of the study, which limits the number of statistically signicant ndings in our study. Future investigators may wish to use a larger sample size. Because cadaveric specimens were used in this study, clinicians should use caution when applying these data to live patients. However, we attempted to maximize the clinical applicability by using fresh-frozen cadavers to replicate tissue consistency and minimize tissue artifact associated with the embalming process, and we used equipment and supplies similar to those we would use in a clinical setting. Furthermore, the average age of the cadaveric specimen donors was 74.1 years, which certainly reects the age group in which knee injections are commonly performed. However, the average body mass index of the cadaveric specimen donors was only 24.75, which is lower than the population average. Therefore the accuracy rates found in this study may not apply to a population in which the body mass index is greater. With the benet of dissection, we did nd interesting anatomic variants. In 4 (20%) of the specimens, we found a plica within the suprapatellar recess, 3 of which were complete (15%) and 1 (5%) that was near complete (Figure 5). Superior plicaes are not uncommon and can present in 20%-55% of individuals [29]. Therefore the presence of a plica was not surprising. However, complete plicaes are relatively rare (2.4%-6.6%), and our sample had a greater than expected number of complete superior plicaes. In patients with a complete superior plica, the injectate would not be allowed to enter the joint. Therefore clinicians should be aware of the potential existence of a superior plica. If US is used to guide the procedure, the clinician should perform a brief pre-procedure diagnostic scan of the suprapatellar recess to try to identify a suprapatellar plica. If one is identied, an US-guided knee injection technique other than the superolateral approach should be used. Furthermore, if during the injection the injectate forms a bolus in the suprapatellar pouch and does not disperse throughout the joint, the presence of a superior plica should be considered and an alternative injection technique should be used.

514

Curtiss et al

SONOGRAPHIC- VS. PALPATION-GUIDED KNEE INJECTION

REFERENCES
1. Felson DT. The epidemiology of knee osteoarthritis: Results from the Framingham Osteoarthritis Study. Semin Arthritis Rheum 1990;20: 42-50. 2. Shimizu M, Higuchi H, Takagishi K, Shinozaki T, Kobayashi T. Clinical and biochemical characteristics after intra-articular injection for the treatment of osteoarthritis of the knee: Prospective randomized study of sodium hyaluronate and corticosteroid. J Orthop Sci 2010;15:51-56. 3. Balazs E, Denlinger, J. Viscosupplementation: A new concept in the treatment of osteoarthritis. J Rheumatol Suppl 1993;39:3-9. 4. Sanchez M, Anitua E, Azofra J, Aguirre JJ, Andia I. Intra-articular injection of an autologous preparation rich in growth factors for the treatment of knee OA: A retrospective cohort study. Clin Exp Rheumatol 2008;26:910-913. 5. Saito M, Takahashi KA, Arai Y, et al. Intraarticular administration of platelet-rich plasma with biodegradable gelatin hydrogel microspheres prevents osteoarthritis progression in the rabbit knee. Clin Exp Rheumatol 2009;27:201-207. 6. Jones A, Regan M, Ledingham J, Pattrick M, Manshire A, Doherty M. Importance of placement of intra-articular steroid injections. BMJ 1993; 307:1329-1330. 7. Sibbitt WL, Peisajovich A, Michael AA, et al. Does sonographic needle guidance affect the clinical outcome of intraarticular injections? J Rheumatol 2009;36:1892-1902. 8. Eustace J, Brophy D, Gibney R, Bresnihan B, FitzGerald O. Comparison of the accuracy of steroid placement with clinical outcome in patients with shoulder symptoms. Ann Rheum Dis 1997;56:59-63. 9. Naredo E, Cabero F, Beneyto P, et al. A randomized comparative study of short term response to blind injection versus sonographic guided injection of local corticosteroids in patients with painful shoulder. J Rheumatol 2004;31:308-314. 10. Jackson D, Evans N, Thomas B. Accuracy of needle placement into the intra-articular space of the knee. J Bone Joint Surg Am 2002;84:15221527. 11. Esenyel C, Demirhan M, Esenyel M, et al. Comparison of four different intra-articular injection sites in the knee: A cadaver study. Knee Surg Sports Traumatol Arthrosc 2007;15:573-577. 12. Toda Y, Tsukimura N. A comparison of intra-articular hyaluronan injection accuracy rates between three approaches based on radiographic severity of knee osteoarthritis. Osteoarthritis Cartilage 2008; 16:980-985. 13. Wind W, Smolinski R. Reliability of common knee injections sites with low-volume injections. J Arthroplasty 2004;19:858-861. 14. Lockman LE. Knee joint injections and aspirations: The triangle technique. Can Fam Phys 2006;52:1403-1404. 15. Bliddal H. Placement of intra-articular injections veried by mini air-arthrography. Ann Rheum Dis 1999;58:641-643. 16. Waddell D, Estey D, Bricker DC, Marsala A. Viscosupplementation under uoroscopic control. Am J Med Sports 2001;3:237-241. 17. Myung J, Lee JW, Lee JY, et al. Usefulness of uoroscopy-guided intraarticular injection of the knee. J Korean Radiol Soc 2007;56:563-567. 18. Carson B, Wong A. Ultrasonographic guidance for injections of local steroids in the native hip. J Ultrasound Med 1999;18:159-160. 19. Qvistgaard E, Kristoffersen H, Terslev L, Danneskoild-Samsoe B, Torp-Pedersen S, Bliddal H. Guidance by ultrasound of intra-articular injections in the knee and hip joints. Osteoarthritis Cartilage 2001;9:512-517. 20. Im S, Lee S, Park Y, Cho S, Kim J. Feasibility of sonography for intra-articular injections in the knee through a medial patellar portal. J Ultrasound Med 2009;28:1465-1470. 21. Koski J. Ultrasound-guided injections in rheumatology. J Rheumatol 2000;27:2131-2138. 22. Smith J, Hurdle M. Ofce-based ultrasound-guided intra-articular hip injection: Technique for physiatric practice. Arch Phys Med Rehabil 2006;87:296-298.

Figure 5. Suprapatellar plica lled with latex. The patellar tendon has been cut and the patella has been reected proximally. P patella; SP suprapatellar plica with latex.

CONCLUSION
In this study, US-guided knee injections with the use of a superolateral approach were 100% accurate, and the accuracy rate of this technique was not inuenced by clinical experience. However, the accuracy of palpation-guided knee injections with the use of a superolateral approach was signicantly inuenced by experience, with the less-experienced investigator demonstrating an accuracy rate of 55%75% and the more experienced investigator demonstrating an accuracy rate of 100%. Therefore when a high degree of accuracy is required and a superolateral approach is considered, our ndings suggest that less-experienced clinicians should consider using US guidance to improve the accuracy of their knee injections. Further research is required to conrm these ndings in the clinical setting and to determine whether the same accuracy rates can be obtained with USguided and palpation-guided knee injections in other populations, such as those with a larger body habitus or signicant joint deformity or when using different injection techniques.

PM&R

Vol. 3, Iss. 6, 2011

515

23. Schweitzer ME, Falk A, Berthoty D, Mitchell M, Resnick D. Knee effusion: Normal distribution of uid. Am J Roentgenol 1992;159:361-363. 24. Agresti A, Coull BA. Approximate is better than exact for interval estimate of binomial proportions. Am Stat 1998;52:119-126. 25. DeFiori E, Rampinelli C, Turco F, Bonello L, Bellomi M. Role of operator experience in ultrasound-guided ne needle aspiration biopsy of the thyroid. Radiol Med 2010;115:612-618. 26. Van Holsbeke C, Daemen A, Yazbek J, et al. Ultrasound experience substantially impacts on diagnostic performance and condence when

adnexal masses are classied using pattern recognition. Gynecol Obstet Invest 2010;69:160-168. 27. Waddell DD, Bricker D. Total knee replacement with Hylan G-F 20 use in patients with grade IV osteoarthritis. J Manag Care Pharm 2007;13:113-121. 28. Lussier A, Cividino A, McFarlane C, et al. Viscosupplementation with hylan for the treatment of osteoarthritis: Findings from clinical practice in Canada. J Rheumatol 1996;23:1579-1585. 29. Dupont JY. Synovial plica of the knee: Controversies and review. Clin Sports Med 1997;16;88-122.