Results of our survey show that, on average, the dose equivalents of corticosteroids used for injecting the various shoulder regions do not differ significantly between orthopaedic surgeons, rheumatologists, and specialty physicians (PCSMs = primary-care sports medicine physicians, and PMRs = physical medicine and rehabilitation physicians). This contradicts one facet of our first hypothesis predicting that significant differences would be found in corticosteroid doses between these groups. Results of our retrospective sample-size and power calculations (see Methods) also suggest that these small differences are not the result of statistical error. Although table 2 provides general guidelines for recommended doses and dose equivalents for common injectable corticosteroids, for most shoulder conditions it is not clear how much corticosteroid/anesthetic is appropriate since uniform guidelines for these injections are not firmly established. This is evident in the broad variations in dose ranges recommended in recent literature for commonly injected locations/conditions (Table 1). The present study shows that there is a wide variety of corticosteroid types used for each injected location for degenerative and/or overuse conditions and reported doses show a broad range within each specialty group. However, only 3–4% of surgeons and PCSMs/PMRs use doses exceeding the "recommended" range for the G-H joint and the S-A bursa. In contrast, 39% of all physicians (41% orthopaedic surgeons, 44% rheumatologists, 29% PCSMs/PMRs) exceeded the recommended dose range for the A-C joint. It is possible that the most commonly used corticosteroids, Depo-Medrol® and Kenalog®, reflect their comparative costs in U.S. dollars (e.g., $5.00 for 40 mg Depo-Medrol®; $6.90 for 40 mg Kenalog®; $6.80 for 6 mg Celestone Soluspan®; $6.00 for 20 mg Aristospan®; $7.60 for 4 mg Decadron®).
In contrast to data for corticosteroid doses, the volumes of local anesthetic used for injecting these painful shoulder conditions varies significantly between the physician groups, corroborating this facet of our first hypothesis. For example, in the perspective of our shoulder specialty practice the amount of local anesthetic typically used by PCSMs/PMRs and rheumatologists seems insufficient, especially for the S-A bursa. This view is based on our use of volumes commensurate with Neer's "impingement test", where 10 cc of local anesthetic (1% Lidocaine) is used to adequately infuse the S-A bursa [20]. Pain relief with provocative maneuvers of the shoulder, indicative of a positive test, is useful for establishing a S-A impingement lesion as the cause of pain [21]. However, most of the physicians (predominantly non-surgeons) used less than half of the 10 cc volume described by Neer. We speculate that the relatively smaller volume of anesthetic (5 cc) used by Kirkley et al. (2002) [22] to rigorously evaluate the Neer impingement test might help to explain why these investigators concluded that this test might be insufficient for predicting the success of surgery [23]. Recent published recommendations for total dose/volume of corticosteroid/anesthetic solutions also typically fall 3–6 cc short of the 11–12 cc that would be administered when corticosteroid is added to the volume of local anesthetic (10 cc) used in the Neer impingement test [3, 4, 24, 25]. Infusing corticosteroid with such 'large' volumes of local anesthetic may also enhance the distribution of corticosteroid throughout the bursa, especially when inflammatory bursa separations or compartmentalization exists [26]. In our shoulder-specialty practice we consider 'larger' volumes of local anesthetic, such as those described by Neer [20], as an essential component for these injections [23]. In turn, results of the present study showing variations between specialists in anesthetic doses suggest that surgeons (who use significantly larger anesthetic volumes) emphasize determining the percentage of pain attributable to the injected region as a tool for surgical planning. However, it is also plausible that, when compared to surgeons, non-surgeons inject lower volumes of local anesthetics because they have more profound knowledge of potential cardiovascular side effects resulting from the inadvertent intra-vascular injection of these agents. For example, there are reports of hypotension, bradycardia, and cardiac arrest resulting from the intra-vascular injection of Bupivicaine and Lidocaine [27–31].
Effective uniform guidelines for corticosteroid injections must consider the possibility that doses exceeding recommended ranges could cause additional systemic and local side effects. Deleterious consequences and other sequelae of corticosteroid injections are frequently a result of chronic use, and may be more prevalent if the corticosteroid spreads to adjacent tissues [8, 32, 33]. Consideration for systemic consequences are also important, especially for patients with diseases that cause immunosupression (e.g., diabetes mellitus and rheumatoid arthritis) [4, 16, 17, 34, 35]. Systemic dissemination of corticosteroid can occur after local injection, further reducing endogenous cortisol levels and exacerbating immunosupression [32, 36, 37]. In this perspective it is interesting that only 10% (15/150) of the survey respondents (two of these were rheumatologists) report any modifications in corticosteroid dose or type, or adjustments in glucose monitoring for diabetic patients. In our clinical practice we have our diabetic patients monitor their glucose levels more closely (i.e., every six hours), and have them adjust their medication doses for one-to-two weeks after receiving the corticosteroid injection. Although we have observed that insulin-dependent and non-insulin-dependent diabetic patients can have broad fluctuations in finger-stick blood glucose measurements (exceeding 400 mg/dL) for several days following an S-A or G-H injection, published data on this issue are contradictory [38, 39]. Further studies are therefore warranted for determining whether adjustments of corticosteroid types, doses, and dose intervals are necessary when injecting diabetic patients.
It has been noted in rheumatology literature that fluorinated corticosteroids (e.g., Celestone Phosphate®, Celestone Soluspan®, Decadron®, Kenalog®, and Aristospan®) (Table 2), when compared to non-fluorinated corticosteroids (e.g., Hydeltrasol®, Depo-Medrol®, Hydeltra-TBA®, and Hydrocorticone®), are more highly associated with tendon rupture and subcutaneous atrophy [8]. For this reason, Moore suggests avoiding the use of fluorinated corticosteroids for extraarticular injections [8]. This is because soft tissue injections of these agents can cause significant atrophy of collagenous tissue that can lead to ligament or tendon rupture or subcutaneous calcification and/or atrophy (the atrophy may disappear in 2 to 3 years). Results of our survey suggest that few physicians, especially orthopaedic surgeons, recognize or accept the putative importance of this issue for selecting a particular corticosteroid type for extra-articular injections. This seems evident in results showing that 17% of rheumatologists, 8% of PCSMs/PMRs, and 37% of orthopaedists typically used fluorinated corticosteroids for injecting the biceps tendon sheath, and 10% of rheumatologists, 5% of PCSMs/PMRs, and 34% of orthopaedists use fluorinated corticosteroids for injecting trigger points.
Our survey also queried physicians regarding whether or not they made adjustments in corticosteroid dosing based on patient age. The only notable finding was that less than 1% of all survey respondents injected painful (non-arthritic) A-C and G-H joints in young patients or athletes. This probably reflects the philosophy, common in sports medicine literature, that performing these injections could exacerbate the injury because pain-related perception, which helps to 'protect' the joint by limiting its use, is diminished [32, 40–42]. Similarly, it has been suggested that pain relief from corticosteroid injections into degenerative joints can accelerate the degenerative process by allowing increased use (and further deterioration) of the joint [40, 43]. Although this idea is typically applied to weight-bearing joints, it might also apply to the A-C and G-H joints.
The possibility that exceeding dose ranges for intra-articular injections (as reported by ~37% survey respondents for the A-C joint) can have deleterious effects on articular cartilage is, however, suggested by indirect or anecdotal evidence. For example, injections of high dose (50–100 mg) betamethasone acetate once a week for two-to-four weeks into the knees of four-to-six month old rabbits caused distortion of articular chondrocyte shape and loss of cell organelles, thus affecting the normal production of collagen fibers responsible for articular cartilage strength [44, 45]. Subtle cell distortion was detected after only two doses (necropsy at 28–42 days post-injection). These and other studies showing deleterious effects in growing animals [32], however, might not be applicable when 'high' doses are injected into degenerative joints in humans. For example, Hollander and co-workers have shown little evidence of morbidity of intra-articular corticosteroid injections that have been used for many years as a primary treatment modality for osteoarthritis of the knee [46, 47]. Owen (2001) [32] also notes that the concept of glucocorticoid arthropathy is based largely on anecdotal case reports and subprimate animal studies. By contrast, studies of primate models have shown no long-term adverse effect on cartilage [48]. There is also a well documented report of a 51-year-old women who received 100 glucocorticoid injections [using Hydeltra TBA, Celestone Soluspan, or Kenalog] into each knee during a span of 10 years with no deleterious effects seen on knee radiographs taken before and after these treatments [49]. Additionally, Balch et al. (1977) studied knee radiographs of 65 patients with osteoarthritis or rheumatoid arthritis who received repeated injections extending from four to 15 years [50]. The radiographs of 15 patients showed no deterioration, 38 showed minimal to moderate deterioration, 10 showed marked deterioration, and only two showed gross deterioration. Although twelve of these patients (12/65, 18.5%) appear to have clearly developed what some might call "steroid arthropathy", the results in the majority of these patients do not support the contention that repeated intra-articular injections of corticosteroids into arthritic knees would inevitably lead to rapid joint destruction.
Even in view of these data and observations – which are now early 20 years old – deleterious consequences of intra-articular injections shown in studies of sub-primate animals appear to have swayed most experts toward warning that similar consequences might also occur in degenerative joints in humans. As noted, our survey did not evaluate the dosing intervals typically used by the physicians in each group. Therefore, it is not clear if 'excessive' doses – irrespective of dose intervals – used by some of our surveyed physicians could enhance the progression of arthritis.
Results of correlation analyses showed a tendency toward reduced corticosteroid doses with years-in-practice in the PCSMs/PMRs and rheumatologist groups. The strongest relationship was found in the rheumatologists who showed reduced corticosteroid doses for periscapular trigger points (r = -0.759, p = 0.05). Additional studies would be required to first corroborate these findings and, second, to determine if they reflect temporal differences in training or a trend that occurs with clinical experience. For example, it would be interesting to determine if this trend in trigger points reflects clinical experience that these rheumatologists might have with subcutaneous atrophy caused by injecting corticosteroid injections.
Limitations of our study include its regional focus on physicians in our tri-state referral area of the western United States. Although these results may not apply to broader geographical regions of the United States, in the context of our literature review they do illustrate the lack of uniform injection guidelines in the general literature. Furthermore, our data might not reflect the injection practices of surgeon and non-surgeon peers outside of the United States. Another limitation included the relatively lower response rate of PCSMs/PMRs (56%), when compared to the higher response rates of rheumatologists and orthopaedic surgeons (83% and 63% respectively). Early-responder vs. late-responder analyses did not demonstrate significant differences within the two groups with response rates of <65%. However, given that there were only 11 and 7 responses in these respective groups from the later mailing, it is unlikely that the data in this context is suitably powered for finding any statistically significant differences. Although these data do not rigorously rule-out early-responder vs. late-responder bias, they tend to reduce this possibility as a confounding factor in generalizing the results as representative of the injection practices of the physician groups that were surveyed. Therefore, the current study provides useful baseline data for guiding future studies of physicians in larger geographical areas. In addition to aiding in establishing uniform injection guidelines, our results also indicate the clear need for advanced education with regards to some aspects of the use of corticosteroid/anesthetic injections.
Future surveys that are designed for broader geographical distribution should also consider additional medical conditions or concerns that were not explicitly addressed in the present study. For example, further clarification of clinical practice of the surgeons and non-surgeon groups should include questions regarding whether or not they would consider injecting, or modifying injections, in patients taking anticoagulant medications. For example, although many physicians might consider the concomitant use of Warfarin as a contraindication of these injections, it is not know exactly how prevalent this opinion is and/or if this opinion varies in cases where patients are taking other types of anticoagulant medications (e.g., aspirin or clopidogrel bisulfate).