Impact of Intra-Articular Corticosteroid Injection on Glycemic Control: A Population-Based Cohort Study

In this retrospective cohort study using Mayo Clinic electronic health record (EHR) data, we 1) assessed whether patients experienced a greater-than-expected increase in A1C after IACS injection and 2) quantified the rate of DKA/HHS in the 30 days after IACS. This study was approved by the Mayo Clinic Institutional Review Board, carried out in accordance with the Declaration of Helsinki, and reported in accordance with Strengthening the Reporting of Observational Studies in Epidemiology guidelines for observational cohort studies (21,22).

We identified adults ≥18 years of age from Olmsted County, MN, who received a large-joint IACS injection from 1 January 2012 through 31 December 2018. Large-joint injections were identified using Current Procedural Terminology codes (Supplementary Appendix). Eligible patients also had to have at least one A1C measurement within the 3 months before and 3 months after their IACS injection. Patients were excluded if they had a previous IACS injection within the 18 months before the index IACS injection so their baseline A1C would reflect glycemic control unaffected by previous IACS.

Patient demographics were ascertained from the EHR at the time of the index date (first IACS injection) and included age, sex, race, and ethnicity. A1C values during the 18 months before and after the index date were collected from the EHR. Diabetes diagnoses were established using International Classification of Diseases (ICD) codes (Supplementary Appendix). Comorbidities were extracted from the EHR from the 12 months before the IACS injection using code sets adapted from the Elixhauser comorbidity measure and the Diabetes Complications Severity Index (23,24). Glucose-lowering medications were identified from the active medication list within the EHR at the time of IACS, and patients were categorized as being on no diabetes medication, being treated with noninsulin medications (i.e., metformin, sulfonylurea, thiazolidinedione [TZD], dipeptidyl peptidase 4 [DPP-4] inhibitor, glucagon-like peptide 1 [GLP-1] receptor agonist, sodium–glucose cotransporter 2 [SGLT2] inhibitor, or meglitinide), or being treated with insulin (basal only or basal-bolus therapy with or without noninsulin medications). Patients with a large-joint IACS injection in the 3 months after the index injection were classified as having multiple IACS injections.

The primary outcome of this study was greater- than-expected increase in A1C after IACS injection, defined as an increase ≥0.5% above the expected A1C based on individual regression lines, as described below. A1C measurements were collected using laboratory data from the EHR.

The secondary outcome was rate of DKA/HHS after IACS injection. Potential episodes of DKA/HHS were identified using ICD codes (Supplementary Appendix). An episode of DKA/HHS within 30 days of IACS was considered potentially related to IACS, and chart review was performed by the study team to determine whether the episode was truly DKA/HHS and whether the severe hyperglycemia could be attributed to IACS.

For each patient included in the study, A1C measurements during the 18 months before and after the index IACS injection during the study time frame were used to construct individual linear regression lines of A1C values. These regression lines were used to estimate predicted A1C values after IACS injection. The first A1C value after the index IACS injection (termed the “actual A1C”) was excluded from the regression model. Patients’ actual A1C was compared with the predicted value, and if the actual value was ≥0.5% greater than the predicted value, it was deemed a greater- than-expected increase in A1C.

A multiple logistic regression model was then fitted for the outcome of greater-than-expected increase in A1C and the covariates of age, sex, race (White vs. non-White), diabetes diagnosis, single versus multiple IACS injections, diabetes medications (none vs. noninsulin medications only vs. any insulin therapy), A1C before IACS (<7.0 vs. 7.0–8.0 vs. >8.0%), and number of comorbidities (0–3 vs. 4–6 vs. >6).

Categorical variables were reported using frequencies and percentages, and continuous variables were reported with mean ± SD and range. χ² tests were used to compare categorical variables between groups with and without a greater-than-expected increase in A1C, and Kruskal-Wallis tests were used to compare the differences for continuous variables. Results from the model were reported as odds ratios (ORs), with their 95% CIs and P values. Two-sided P values <0.05 were considered statistically significant. All analyses were performed in SAS, v. 9.4, software (SAS Institute, Cary, NC).

We identified 1,169 patients who received a large-joint IACS injection and were included in the study. Their mean age was 66.1 ± 12.2 years, 52.8% were female, and 92.6% had a diabetes diagnosis before the IACS injection (Table 1). Among all patients, 184 (15.7%) experienced a greater-than-expected increase in A1C (≥0.5% above the predicted A1C) after large-joint IACS injection (Table 1). Overall, mean A1C was 7.3 ± 1.4% before IACS and 7.2 ± 1.4% after IACS. The difference between actual and predicted A1C was 0.0 ± 0.8% (Table 2).

Patients with a greater-than-expected increase in A1C were, on average, 64.8 ± 12.4 years of age, and 95 (51.6%) were female, compared with a mean age of 66.3 ± 12.0 years and 522 (53.0%) female in the group without a greater-than-expected increase in A1C. There was no statistically significant difference in age or sex between the two groups. The group with a greater-than-expected increase in A1C had a higher proportion of patients with a formal diabetes diagnosis than the group without (97.3 vs. 91.8%, P = 0.009). Overall, 6.8% of patients had multiple IACS injections, and the proportion of patients with multiple IACS injections was not different between the two groups. The group with a greater-than-expected increase in A1C had higher percentages of patients with substance abuse (7.1 vs. 3.5%), cerebrovascular disease (9.2 vs. 5.0%), and previous episodes of hypoglycemia (2.2 vs. 0.5%) compared with the group without a greater-than- expected increase in A1C. Patients with a greater-than-expected increase in A1C also had higher rates of sulfonylurea, SGLT2 inhibitor, and basal insulin therapy than those without a greater-than-expected A1C increase (Table 1). Mean A1C before IACS was 8.0% among those with a greater-than-expected A1C increase and 7.1% in those without a greater-than-expected increase (P <0.001). On average, A1C rose 1.2% (range 0.5–6.4%) in the group with a greater-than-expected increase compared with an average decrease of 0.2% (range −5.5% to 0.5%) in the group without a greater-than-expected increase (Table 2).

In multivariable regression analysis, baseline A1C was the only factor associated with a greater-than-expected increase in A1C after IACS injection. The odds of a greater-than-expected increase in A1C in patients with a baseline A1C between 7.0 and 8.0% were 1.84 (95% CI 1.08–3.13) compared with those with a baseline A1C <7.0%. In patients with a baseline A1C >8.0%, the odds of a greater-than-expected increase in A1C after IACS were 4.79 (95% CI 2.83–8.13) (Table 3).

Among the 1,169 patients who received IACS injections during the study time frame, five were identified as having possible DKA/HHS after IACS. Upon chart review, only one patient had DKA attributable to IACS. This patient was a 60-year-old woman with type 1 diabetes who had received 80 mg triamcinolone into a knee joint and had a pacemaker lead removed on the same day. Her A1C on the day of injection was 9.4%. She had medical comorbidities of squamous cell lung cancer, chronic kidney disease, hypertension, chronic obstructive pulmonary disease, diabetic neuropathy, peripheral vascular disease, and hypoglycemic seizures and had had at least three previous episodes of DKA, with the most recent episode occurring 1 year earlier.

IACS injections are commonly administered to people with diabetes who would benefit from nonoperative symptom management of degenerative and inflammatory joint conditions. This study carefully characterized the effects of IACS on glycemic control in a well-characterized population-based sample (25). We were reassured to find that the majority of patients in our study did not have a greater-than-expected increase in A1C after large-joint IACS, suggesting that IACS injections do not substantially worsen diabetes control or accelerate progression of diabetes in most people who receive them.

Although most patients did not have worsened glycemic control after IACS, nearly 16% did experience an increase in A1C after large-joint IACS above what would be expected based on their disease history and progression. Thus, it is important to establish what clinical factors may predict a greater increase in A1C after IACS administration. In our population, baseline A1C was the most important factor in predicting worsening glycemic control after IACS, with those with an A1C >8.0% having the highest odds of having a greater-than-expected increase in A1C. These findings are consistent with previous smaller studies showing greater elevations in blood glucose after IACS in patients with a higher baseline A1C (13,26) and suggest that baseline A1C should be considered when prescribing IACS injections.

Number of comorbidities was not predictive of a greater-than-expected increase in A1C. However, the diagnoses of diabetes, substance abuse, and cerebrovascular disease and history of hypoglycemia were more common in the group with a greater-than-expected A1C increase after IACS. Additionally, prescriptions for a sulfonylurea, SGLT2 inhibitor, and basal insulin were more prevalent among those with a greater- than-expected A1C increase. Additional caution may be taken when prescribing IACS in patients with these characteristics, and future studies are needed to elucidate the specific risks associated with individual comorbidities and medications.

In our analysis of episodes of DKA/HHS after IACS, only one patient of 1,169 had an episode of severe hyperglycemia that was attributable to IACS. The low rates of DKA/HHS are reassuring and in keeping with previous reports (13,14,27). Caution should be taken in patients with poorly controlled diabetes and a history of DKA, but concern about the possibility of dangerous hyperglycemia should not delay IACS administration in most cases. High-risk patients should be advised to monitor their blood glucose closely and be educated on sick-day guidelines for hyperglycemia self-management.

Our study must be considered in the context of its limitations. This was a retrospective study and therefore could not establish causation. Furthermore, there was no control group for comparison, which limited our ability to compare the effect of IACS to the natural history of A1C progression. The population was also from a single health care system, with few patients of racial and ethnic minority backgrounds. We also did not have information on corticosteroid formulation, total steroid dose, or specific location of IACS injection. Differences in steroid solubility and joint surface area for corticosteroid absorption could be important factors in determining specific risks of hyperglycemic outcomes and should be an area of further investigation (28,29). Similarly, we do not have information on changes in glucose-lowering medications, which also affect A1C control and could alter the rate of change in A1C in either direction. This study was also not designed to evaluate hyperglycemic effects in the context of other injectable or systemic corticosteroids, which could have biased the results toward no difference.

Strengths of the study include its size and population-based capture of IACS data. Our study included 1,169 patients, compared with much smaller cohorts of <60 patients in most other studies examining the hyperglycemic effects of IACS (9–14). This large patient population allowed for better evaluation of risk factors for hyperglycemia and enabled us to monitor for less prevalent hyperglycemic events (i.e., DKA/HHS). Additionally, more patients in our population had an A1C >8.0% compared with previous studies (14–17), allowing a better evaluation of this key risk factor for hyperglycemia. Our study also evaluated the longer-term glycemic effects of IACS by following change in A1C rather than blood glucose measurements. Changes in A1C are more clinically relevant to long-term diabetes management than temporary blood glucose elevations because prolonged hyperglycemia increases the risk of diabetes-related complications.

Our findings improve the understanding of the glycemic impact of IACS by characterizing and detailing potential risk factors for a greater-than-expected increase in A1C. Currently, there are no guidelines for the use of IACS injections in patients with suboptimally controlled diabetes. These injections are important procedures for pain relief and should not be delayed. They are often preferable to alternative pain management options such as surgery or opioid pain medications. However, it is important to consider the glycemic implications when prescribing IACS for pain relief. If patients’ A1C is >8.0%, clinicians should consider monitoring them closely after IACS injection, referring them to diabetes self-management education and support, and engaging the support of clinical team members, including nurses, pharmacists, and certified diabetes care and education specialists, to help patients monitor their glucose levels and adjust their treatment regimens to better control any resulting hyperglycemia.

This population-based study examined increase in A1C as a marker for longer-term glycemic control after large-joint IACS. The results suggest that, while most people will not have adverse long-term glycemic outcomes after IACS, those with elevated A1C at baseline are at risk for further increase in their A1C, as was seen in 16% of patients receiving large-joint IACS injections. Baseline A1C should therefore be considered when prescribing large-joint IACS, with a focus on optimizing glucose control and close post-injection glucose monitoring for patients with elevated A1C. These results will need to be confirmed in larger prospective trials, in more heterogeneous populations, and with control groups for comparison.

This publication was made possible by Clinical and Translational Science Award grant UL1 TR002377 from the National Center for Advancing Translational Sciences, a component of the National Institutes of Health (NIH). It is also funded by the NIH’s National Institute of Diabetes and Digestive and Kidney Diseases grant K23DK114497 (R.G.M.). T.T.S. received research support through the Mayo Clinic Department of Medicine Catalyst for Advancing in Academics Award. This publication’s contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIH.

No potential conflicts of interest relevant to this article were reported.

T.T.S. researched data and wrote the manuscript. L.S.G. and R.G.M. researched data and reviewed/edited the manuscript. K.M.F. performed statistical analysis, contributed to the methods, and reviewed/edited the manuscript. T.T.S. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.