Upper vs. lower extremity: Does the site of steroid injection have a different effect on blood glucose levels in patients with diabetes?

A clinical decision report appraising Twu J, Patel N, Wolf JM, Conti Mica M. Impact of variation of corticosteroid dose, injection site, and multiple injections on blood glucose measurement in diabetic patients. J Hand Surg Am. 2018;43(8):738744. https://doi.org/10.1016/j.jhsa.2018.06.005 for a patient with diabetes and osteoarthritis.


Related Literature
A review of UpToDate was initially performed to evaluate the current management of patients with moderate to severe OA, as well as the comorbidities that impact the management. 1,2 Currently, the management of OA involves both pharmaceutical and nonpharmaceutical modalities, aimed around reducing inflammation and physical therapy. While intra-articular glucocorticoid injections are not routinely used for patients with knee OA, their use can provide some short-term relief to patients with moderate to severe OA. 1 For patients with diabetes, glucocorticoid injections can produce a transient hyperglycemia, and therefore it is advised that patients monitor their blood glucose for a few days post-injection. 2 A subsequent search of PubMed was performed using the terms "diabetic" AND "joint injection" AND "blood glucose" returned 33 results. The titles and abstracts were reviewed to find publications relevant to the clinical question. This further refined the results to 15 articles. Of these results, 2 were systematic reviews and 5 were clinical trials. The titles and abstracts of these 21 results were used to determine relevance in relation to the clinical question.
One article discussed the impact of intra-articular betamethasone injections for diabetic patients. 3 However this study focused on the impact on insulin resistance not blood glucose levels and therefore was omitted. Another looked at the impact of methylprednisolone acetate on fructosamine levels, but again was omitted as it did not pertain directly to the clinical question. 4 A publication by Zufferey et al. studied the effects of epidural steroid injections on blood glucose in patients with diabetes mellitus and found that there was no significant increase in blood glucose levels post-intervention. 5 This paper found conflicting results with one published by Younes et al., but brought up the discussion of how glycemic effects could differ based on the type of steroid used (soluble versus depot type). 6 However, because these papers addressed the use of epidural injections, not intra-articular, they were deemed less relevant to the clinical question.
Two systematic reviews discussed the systemic effects of intra-articular glucocorticoid injections in diabetic patients. [7][8] Both provided insight as to other effects that may need to be considered when a glucocorticoid injection is offered to a diabetic, but were considered too broad in relation to the clinical question.
There are many articles are available discussing the changes in blood glucose after shoulder injections, where the overall consensus is that there is no significant increase in blood glucose levels. [9][10][11] There was also an article that noted a significant increase in blood glucose after knee injections. 12 While these articles could all be considered relevant to the clinical question, there was one study that directly compared upper versus lower extremity injections and their effects on blood glucose. Utilizing the Twu et al. paper, the comparison of upper versus lower extremity and effect on blood glucose could be directly answered, and therefore was the paper chosen to analyze. 13 Given the above sources discussed, this body of evidence is a Grade-B Strength of Recommendation based on the SORT criteria.

Critical Appraisal
The article by Twu et al. is a prospective cohort study to monitor the effects on blood glucose of 60 patients with diabetes mellitus who were offered a corticosteroid injection for either an upper or lower joint injection. Patients had to have a documented diagnosis of diabetes mellitus, be above the age of 18, able to complete glucose measurements, and not have received any injections in the past 3 months to be included in the study. Once enrolled, patients received standard injections of triamcinolone based on their assigned injection site, and were given a log to record one fasting and one 2-hour postprandial blood glucose measurement for 14 days post-injection. Upon completion of the 14 days, researchers contacted study participants to retrieve measurement data. Based on this study design, this article is considered to have a SORT evidence level of 3.  Table 1 in the article outlines the demographics and baseline characteristics of patients who received different injections. Patients were separated into 3 groups based on the type of injection they were receiving; a single upper extremity injection (n=32), a single lower extremity injection (n=17), or multiple injections (n=11). These patients were approached to join the study based on their diagnosis and treatment offered. Because of this, the study may present with selection bias. As well, the groups were assigned based on documented injection needed, therefore there was no randomization amongst study groups.
It can also be noted that there were some baseline differences between groups. In particular, the average HbA1C varied from 7.65 for patients receiving an upper extremity injection, 7.45 for lower extremity, and 7.6 for multiple injections. Our patient was a 63 year old male receiving a lower extremity injection, with an HbA1C of 7.2. This would indicate that his diabetes was slightly better controlled than then average patient receiving the same injection as him in the study.
Each group received a separate experimental intervention depending on their study group. Since each type of injection requires a specific amount of steroid, patients receiving an upper extremity injection received less steroid than someone requiring a lower extremity injection. This difference in steroid dose could have accounted for some of the differences seen in blood glucose changes post-injection. However, aside from the differences in intervention, all study groups followed the same experimental protocol.
A mixed-model covariant analysis was used to interpret the significance of the changes in fasting and postprandial blood glucose, which accounted for day-to-day measurements, corticosteroid dose, single versus multiple injections, BMI, insulin use, and HbA1C. This allowed for the use of both fixed effects, like how groups differ between a set of treatments, and random effects, such as the variability among study subjects. The entire cohort was analysed together, followed by analysis of each group separately.
The outcome was based on a comparison of days 1-7 compared to days 10-14, with the later days considered "baseline" on the assumption that a change in blood glucose would only occur within the first week. Changes in blood glucose were the highest on day 1-elevated by 35 mg/dl, a clinically meaningless difference. In fact, when reviewing the methodology and the outcomes chosen by the researchers, this article seems to report trivial findings. The usefulness of the evidence is that there was not a bad outcome in any patient, which could be used to reassure Mr. Carpenter.

Clinical Application
Glucocorticoid injections are an effective treatment for a variety of upper and lower extremity joint conditions such as osteoarthritis. However, these steroids have been shown to affect glucose metabolism, and cause abnormal blood glucose measurements in diabetics. This raises the concern of many different glucose-related complications, such as diabetic ketoacidosis, that could arise if diabetics receive an intra-articular injection. The findings by Twu et al. suggest that the location of injection (upper versus lower extremity) does play a role in the differences of post-injection blood glucose measurements for patients with diabetes mellitus. While all patients did see some increase in both fasting and postprandial blood glucose levels up to 72 hours post-injection, this increase was not significant for patients who received upper extremity injections or multiple injections. However, patients who received lower extremity injections were found to have a statistically significant and clinically meaningless increase in fasting blood glucose, with no impact on postprandial blood glucose. The study also found that as HbA1C increased, more significant increases in blood glucose were also seen.
For our patient Mr. Carpenter, this study suggests that he should indeed monitor his blood glucose levels more closely upon receiving a glucocorticoid injection in his knee, as there may be a significant increase in the weeks post-injection. This study could also be used to explain why Mr. Carpenter did not raise any concern when he received a similar injection in his shoulder many years ago. It should be noted that Mr. Carpenter's demographics did not match up perfectly with the study participants considering he had a lower HbA1C of 7.2, and is Caucasian (90% of study participants were of black ethnicity), and therefore this study cannot truly predict his blood glucose response to glucocorticoid injections. Nonetheless, we thoroughly counselled Mr. Carpenter on the importance of blood glucose control, and that closer monitoring of his blood glucose post-injection was encouraged. In practicality, his dietary consumption is a far greater risk for diabetic control, so we took the opportunity to counsel him on coping strategies to manage calorie intake during the upcoming festivities. Considering how prevalent