Prospective Evaluation of Pain Flares and Time Until Pain Relief Following Musculoskeletal Corticosteroid Injections

Abstract

Background:

Corticosteroid injections are used ubiquitously within musculoskeletal medicine. One of the most common side effects is a postinjection pain flare, though little is known regarding this phenomenon.

Hypothesis:

Some risk factors are related to postinjection pain flare following an ultrasound-guided corticosteroid injection.

Study Design:

Prospective clinical research study.

Level of Evidence:

Level 2.

Methods:

Patients undergoing ultrasound-guided corticosteroid injections in an academic orthopaedic and sports medicine clinic were approached to participate. Patients completed a survey immediately following their injection and again 2 weeks later, asking them about their pain and side effects. A postinjection pain flare was defined as an increase in pain, as defined by the patient.

Results:

A total of 140 patients completed the entirety of the study, with 29 (20.7%) patients reporting a flare of pain. There was a significant effect of younger age on the development of a pain flare after the injection, estimated as 5.5% decreased odds of developing a flare per year of age (P < 0.01). Gender, injection location, body mass index (BMI), preinjection pain, and corticosteroid type had no contributing effect. When patients obtained relief following the corticosteroid injection, 60.4% had improved pain within 3 days, whereas over 93.7% obtained relief within a week.

Conclusion:

Pain flares seem to affect approximately 1 in 5 patients. With increasing age, the likelihood of postinjection pain flare becomes less likely. Sex, injection location, BMI, preinjection pain, and corticosteroid type do not seem to significantly relate to an increase in pain following injection.

Clinical Relevance:

Corticosteroid injections are common procedures in the orthopaedic and sports medicine settings. Younger patients can be counseled on the higher likelihood of a pain flare following a corticosteroid injection.

Corticosteroid injections have demonstrated pain relief in the treatment of a multitude of musculoskeletal pathologies, such as osteoarthritis, tendinopathy, trigger finger, tenosynovitis, and adhesive capsulitis.^{4 -6} Given the ubiquity of corticosteroid injections in modern musculoskeletal practice ¹⁸ within the realm of musculoskeletal medicine, it naturally follows that a variety of side effects have been reported. ²¹ Prior studies have examined the incidence, intensity, and quality of these side effects.^2,11-14

A flare of pain is a common reaction seen in clinical practice before the corticosteroid starts providing longer-term pain relief. Many theories exist as to the cause, including needle trauma, inadvertent placement, anesthetic toxicity, initial enhanced immune response, and corticosteroid crystallization.^1,17,20 There is little in the literature reporting on who will experience a pain flare following a corticosteroid injection, even though the incidence of postinjection pain flares range in incidence from 33% to 50%.^8,10,22 To the best of the authors’ knowledge, no studies have identified a patient population at increased risk of experiencing postinjection pain flares.

In this study, we aimed to identify which type of patient is at greatest risk of experiencing a postinjection pain flare. Specifically, we sought to identify which preinjection patient-dependent factors most likely predispose patients to a postinjection pain flare (transient increase in pain) following an ultrasound-guided corticosteroid injection.

Methods

This study was approved by the primary investigator’s institutional review board. Patients undergoing ultrasound-guided corticosteroid injections in an academic orthopaedic and sports medicine clinic were approached to participate in this prospective study. Any patient receiving a therapeutic corticosteroid injection was eligible, with the following exclusion criteria: previous injection or oral corticosteroid use within the past 3 months, previous surgery in the area of the injection within the past year, and inability to adequately complete surveys (eg, dementia, psychiatric disorders, or inability to access emails).

Each injection was performed by a board-certified sports medicine physician with at least 5 years of experience with ultrasound and injections. All injections involved use of lidocaine for the initial local anesthetic, and a corticosteroid (usually mixed with an anesthetic) for the actual injection. Injections were performed with a 22- to 27-gauge needle, depending on the injection. All patients completed a survey immediately following their injection. The survey included a series of demographic questions, asking them about their pain ranging from 0 to 10 (Numeric Pain Rating Scale ⁷ ). At that time, they were instructed to monitor their pain and particularly note when their pain improved over the next 2 weeks. They received a follow-up survey at the 2-week time point by email, asking them about their pain and side effects.

A postinjection pain flare was allowed, to be subjective. Prior literature has defined it as a 2-point increase on the visual analog scale (VAS), ⁸ which was based on the minimal clinically important difference in pain for the numeric pain rating. In this study, however, we chose to exclude the defined value of 2 numeric points when asking patients about a flare, instead allowing the patient to decide whether it was clinically important or not.

Statistical analysis was performed with Stata Version 17.0 (StataCorp). Comparison of duration until pain relief between patients with pain flares and those without was performed, using the Wilcoxon-Mann-Whitney rank-sum test with the calculations of an exact P value, because of the non-normal distributions of the variable. Logistic regression analysis (outcome variable = dichotomous pain flare) was performed with the following independent variables: sex, age, injection type (joint [baseline], tendon, bursa, perineural), body mass index (BMI), preinjection pain level, and corticosteroid type (methylprednisolone [baseline], triamcinolone). Further, a linear regression model was used to examine the associations of the above named independent variables with amount of time until pain relief (= continuous outcome variable), with the robust or sandwich estimator of variance^15,23 to account for the non-normal distribution of the outcome variable. Statistical significance was set at P < 0.05.

Results

A total of 160 participants were included in the study; Table 1 lists demographic characteristics of the patients. Twenty patients did not complete the follow-up survey, resulting in an 87.5% response rate. Of the remaining 140 patients, 111 (79.3%) reported pain improvement at 2 weeks postinjection.

Table 1.

Demographic characteristics of patients included in the study and information about injections, N = 160

		N	%	Mean	SD
Age		160		52.4	14.9
BMI		160		29.1	6.4
Pain (0-10)		160		5.0	2.3
Sex	Female	90	56.6
	Male	69	43.1
Trainee involvement	Yes	129	80.6
	No	19	11.9
	Unknown	12	7.5
Steroid type	Methylprednisolone	127	79.3
	Triamcinolone	26	16.3
	Dexamethasone	2	1.3
Dose (excluding dexamethasone)	10 mg	6	3.8
	15 mg	3	1.9
	20 mg	16	10.0
	30 mg	20	12.5
	40 mg	98	61.3
	80 mg	9	5.6
Most common diagnoses	Osteoarthritis	84	52.5
	Tendinopathy	27	16.9
	Postsurgical pain	7	4.4
	Adhesive capsulitis	7	4.4
Kellgren-Lawrence grade for osteoarthritis	0	6	7.1
	1	19	22.6
	2	25	29.8
	3	27	32.1
	4	8	9.5
Pain relief by 2 weeks	Yes	111	69.4
	No	29	18.1

As indicated by Table 2, most injections were performed on large joints or bursae. The most common of these were glenohumeral (16.9% of all injections), intra-articular knee (13.1%), tibiotalar (13.1%), hip (11.3%), and subacromial bursa (8.1%).

Table 2.

Types of injections included in study

Injection type	n	Percentage
Large joint/bursa injection	89	55.6
Intermediate joint/bursa injection	33	20.6
Small joint/bursa injection	19	11.9
Tendon sheath injection	9	5.6
Nerve block	6	3.8
Other	4	2.5
Total	160	100

Most patients reported initiation of their pain relief within a few days of injection, as outlined in Figure 1; 50% had obtained relief by 3 days, while 95% of patients had obtained relief by 8 days. Patients who reported a flare of pain (n = 29 of 140, 20.7%) averaged 4.54 days until relief, whereas those without a flare took 3.10 days (P = 0.01). In addition, those who reported a pain flare were as likely to have pain relief by 2 weeks as those who did not report a flare of pain (70.8% vs 81.0%; P = 0.26).

Figure 1.

Days until relief of pain (n = 111). Bars represent the cumulative number of patients who had obtained pain relief by that day.

Logistic regression modeling (full results in Table 3a) demonstrated a significant effect of age on the development of a pain flare after the injection, estimated as 5.5% decreased odds of developing a flare per year of age (P = 0.00). In other words, younger patients were more likely to experience a pain flare; with each passing year, older patients decreased their odds of obtaining a pain flare by 5.5%. Sex, injection location, BMI, preinjection pain, and corticosteroid type all did not significantly relate to an increase in pain (P > 0.05). With respect to time until pain relief, no significant demographic or injection-related factors were identified (Table 3b).

Table 3(a).

Associations with development of pain flair ^a

	OR [95% CI]	P
Sex	1.064 [0.418, 2.708]	0.90
Age	0.945 [0.912, 0.979]	0.00*
Joint (baseline)	1
Tendon	1.714 [0.447, 6.575]	0.43
Bursa	0.861 [0.162, 4.564]	0.86
Nerve	(N/A)
BMI	0.995 [0.923, 1.072]	0.90
Preinjection pain level	0.929 [0.754, 1.144]	0.49
Methylprednisolone (baseline)	1
Triamcinolone	0.263 [0.031, 2.215]	0.22

BMI, body mass index; N/A, not applicable.

^an = 127 patients who provided complete follow-up data at 2 weeks. An odds ratio (OR) identifies the increased or decreased odds of developing a pain flare by each independent variable, while adjusting for the other variables. For example, a 1-year increase in age was associated with 5.5% lower odds (1.000 – 0.945 = 0.055) of developing a pain flare, whereas the other variables held constant (P = 0.00).

^*Bolded numerals indicate statistical significance.

Table 3(b).

Associations with amount of time until pain relief ^a

	Beta coefficient [95% CI]	P
Sex	0.457 [−0.578, 1.492]	0.38
Age	−0.028 [−0.068, 0.011]	0.16
Joint (baseline)	0
Tendon	0.379 [−1.447, 2.205]	0.68
Bursa	−0.713 [−1.960, 0.533]	0.26
BMI	−0.078 [−0.158, 0.003]	0.06
Preinjection pain level	0.197 [−0.087, 0.480]	0.17
Methylprednisolone (baseline)	0
Triamcinolone	−0.942 [−2.352, 0.467]	0.19

^an = 98 patients with complete follow-up data at 2 weeks who reported some improvement in their pain. The beta coefficient describes the change in the amount of time relief related to each independent variable, while adjusting for the other variables. For example, there was a trend of a 1-unit increase in body mass index (BMI) being associated with 0.078 days faster relief (P = 0.06).

Discussion

This prospective study of a variety of corticosteroid injections demonstrated that subsequent pain flares are common. Approximately 1 in 5 patients reported a pain flare; these patients noted prolonged time until relief of their pain, and younger patients were more likely to experience a flare. Several factors were not related to these pain flares: sex, the musculoskeletal structure being injected, BMI, preinjection pain level, or type of corticosteroid.

A few studies have previously investigated the incidence of postinjection pain flares and symptomatic improvement following injection, as well as location of injection and incidence of postinjection pain flares. Fawi et al ⁸ reported that 35.3% of patients receiving a shoulder corticosteroid injection experienced postinjection pain, with a mean duration of symptoms of 3.9 days, but this did not seem to have any detrimental effect on overall outcome. Their duration was similar to those reported in our study. Goldfarb et al ¹⁰ evaluated the incidence of postinjection pain flares following extra-articular steroid injections, comparing a standard-injection group with a pH balanced–injection group. They reported a postinjection pain flare in 33% of patients, with no significant difference among standard versus pH-balanced injection groups. In addition, Wang et al ²² examined the temporal pain response following corticosteroid injections to the hand and elbow. They reported a 50% postinjection pain flare incidence, but most symptoms improved by day 2 status postinjection. These 3 studies all had a larger incidence of pain flares. This may be related to a few factors. First, the other studies did not use ultrasound guidance, which has been demonstrated to improve accuracy of injection placement ⁹ ; injections external to the source of pain may cause a transient flare of pain. Second, these studies looked at more focused areas (shoulder, ⁸ hand tendons, ¹⁰ and hand/elbow ²² ): these upper extremity areas may be more prone to flares than those seen in the lower extremity, which made up a subset of our injections. We did not explicitly evaluate this in our regression model, due to the large number of areas injected, which, even if categorized, would result in a lower statistical power. Third, the other studies prospectively evaluated patients over each day; recall bias could have potentially played a part in our patients’ recollection of their pain. Finally, a pain flare was evaluated using a different methodology. We chose to use a subjective value because we felt it was the most clinically applicable. It is possible the other studies had more patients with mild increases in their pain that our patients did not feel were clinically “worse.”

To the best of our knowledge, no other studies have matched our finding that younger age was more likely to demonstrate increased odds for a pain flare, notwithstanding its small effect size. It is likely the subjective nature of our evaluation of pain flares may have played a role: age is related to pain perception, ¹⁹ and younger patients may have felt the adverse changes in pain to be more clinically relevant than older patients. Younger patients may also have had different musculoskeletal conditions necessitating a corticosteroid injection; we did not categorize all suspected diagnoses, but this could play a role. Last, there may be a more robust inflammatory response to corticosteroid injection in younger patients. ³

A second finding of our study was the length of time until patients reported symptomatic improvement after their injection. Over half reported symptomatic improvement by 3 days postinjection, whereas 95% of patients reported symptomatic improvement by 8 days postinjection. Though other studies have not explicitly looked at a categorically improved versus not improved variable, our findings appear similar to the findings seen in other studies.^8,10,22 Not surprisingly, those patients reporting a postinjection pain flare took, on average, approximately 1.5 days longer to note symptomatic improvement (4.54 days compared with 3.10 days in the non–pain flare group; P = 0.01), which is also in line with previous literature. ⁸ No significant demographic or injection-related factors were found related to duration of time until pain relief.

There are several significant limitations to this study. First, the sample size was limited to 160 participants, and 12.5% failed to respond. Though still a relatively high response rate, selection bias likely played a role. Second, recall bias may have occurred, as we gave patients a survey at 2 weeks. However, we attempted to limit this bias by discussing the study with patients beforehand and asking them to monitor their pain timeline. Third, postinjection pain data were only tracked for 14-day status after corticosteroid injection; it is unlikely that some of the 29 patients who felt no relief may have eventually felt relief after 2 weeks. Fourth, there is a correlation between the pain associated with the injection and the level of pain experienced following the injection. ¹⁶ This study did not take the pain of the procedure into account. Fifth, we did not provide a previously described definition for an increase in pain. Prior literature has suggested a VAS increase of 2 points as being clinically important.^8,10 We could have added this to the questionnaire as well, to identify the level of pain flare that the patients experienced. Finally, we included participants with a variety of musculoskeletal conditions and locations of pain, in addition to providing a variety of injections. Though this may increase the generalizability of the study, it fails to address potential location-based or pathology-based issues related to pain flares or duration of time until pain relief.

Conclusion

When patients obtain relief following corticosteroid and anesthetic injections, half will obtain it within 3 days, whereas over 90% will obtain it within a week. Pain flares are also common and occur in about 1 in 5 patients. With increasing age, postinjection pain flare becomes less likely. If a postinjection pain flare does occur, it takes, on average, 1.5 days longer to note symptomatic improvement compared with those who do not experience a postinjection pain flare.