Sources of Sacroiliac Region Pain: Insights Gained From a Study Comparing Standard Intra-Articular Injection With a Technique Combining Intra- and Peri-Articular Injection

Methods 

Billing records for a single practitioner were reviewed. A list of all patients who had undergone sacroiliac injection from 2002 through 2006 was created. The patients' charts were reviewed in chronologic order, and patients who met inclusion and exclusion criteria were enrolled sequentially into the study until the study included 40 patients who had undergone standard sacroiliac injection and 80 patients who had undergone the new modified procedure. Patients were included if they reported pain in the low back below L4 and buttock, thigh, calf, or groin pain.28 All patients had also failed to improve with physical therapy consisting of any or all of the following: manual manipulation, myofascial release of the lumbopelvic area, pelvic stabilization exercise, and bracing. If potential subjects had a documented disk herniation or lumbar stenosis graded as worse than mild, they had to be unresponsive to a transforaminal epidural injection directed at the root corresponding to their symptom distribution. Presence of moderate stenosis was determined by the attending neuroradiologists' official reports. Consensus definitions for moderate stenosis were lateral recess anteroposterior diameter measuring 3mm or less, or central canal anteroposterior diameter measuring 9mm or less. There was no consensus definition for foraminal stenosis. If the patients had received facet injections, they had to experience minimal or no pain relief.

Patients were excluded for the following reasons. The patient's chart did not include a full database of the potentially confounding variables the study was examining (table 1).5, 8, 11, 29 Patients increased their pain medication use by more than 15% over the course of the 3-month follow-up period. Patients were instructed that medications could be increased for a period of 5 to 7 days postprocedure to address postprocedure pain, but afterward all attempts should be made to reduce medication use to prior usage levels or below. Patients were told exceptions could be made for a serious flare-up of a new pain that they felt would not spontaneously resolve. Patients with psychiatric impairment severe enough to impair their ability to interpret the outcome measurement tools were excluded. Any patient experiencing anesthesia below the buttock or leg weakness postinjection was excluded because this was presumed to be a nonspecific block.

Table 1. Sample Demographics and Symptom Distributions

	Variable	SIJ Injection (n=40)	SI+S1-3 Injection (n=80)
	Age (y)	56.33±10.61	49.86±13.32
	Duration of symptom⁎ (mo)	36.1±27.91	47.1±42.72
	Mean initial VAS pain scores	7.88	7.95
	Sex (male/female)	10/30	24/56
	Unemployment	72.5	66
	Obesity	20	26
	Depression	50	41
	Diabetes	10	7.5
	Smoking†	35	38.8
	Oxycodone use >120mg/d	25	27.5
	Oxycodone use >60mg/d	30	28.75
	Oxycodone use >30mg/d	65	57.5
	Lumbar fusion	22.5	22.5
	Stenosis	35	28.8
	Bilaterality	32.5	45
	Traumatic	20	33.8
	Pain pattern 1	100	100
	Pain pattern 2	42.5	46
	Pain pattern 3	32.5	21.25
	Provocative sit	70	67.5
	Provocation by standing†	80	88.8
	Provocation by walking†	95	85
	Provocation by sit to stand‡	77.5	42.5
	Provocation by stairs†	77.5	67.5
	Provocation by side lay	17.5	27.5
	Positive SLR	20	18.8
	Provocation by hip AFPS	92.5	80
	PSIS tenderness	87.5	88.8

NOTE. Values are mean ± SD, percent, or as otherwise indicated. Pain pattern 1 includes the back, buttock, and thigh; pain pattern 2 is below the knees; pain pattern 3 includes the groin area. Abbreviations: Hip AFPS, hip adduction, flexion, posterior shear; SI+S1-3, combined sacroiliac and S1-3 injection; SLR, straight-leg raise.

⁎ Variable correlated with treatment outcome at 3 months for sacroiliac group. † Variable correlated with treatment outcome at 3 months for combined sacroiliac and S1-3 injection group. ‡ Significant difference between the treatment groups in frequency of this variable.

A total of 167 records were reviewed to obtain the 120 study subjects. Only 3 subjects were excluded because of increases in narcotic dosing (1 in the standard injection group, 2 in the new modified injection group). All 3 increased their medications because of problems unrelated to their sacroiliac region. Two had exacerbations of neck and shoulder pain, and 1 had new-onset headaches. Two were excluded for leg anesthesia, both in the modified injection group. Thirty-two subjects were excluded because of inadequate databases. Eight subjects were excluded as a result of previous pain relief by epidural injections. Two patients were excluded retrospectively because, during review of these patients' cases in regularly scheduled case conferences, our attending psychologist suggested they had Axis II disorders severe enough to inhibit their ability to use our outcome measurement tools appropriately. One of these patients was in the standard injection group and 1 in the new modified injection group.

Intra-articular sacroiliac injection was done according to the technique described by Fortin et al.27 The extra-articular injection was done by modifying slightly the technique described by Yin et al.30 The injection site is targeted by obtaining an oblique fluoroscopic view of the PSIS that aligns the medial edge of the PSIS with the medial edge of the SIJ line (about 15°–20° oblique). A 22-gauge, 3.5-inch to 5-inch curved spinal needle is advanced in line with the fluoroscopic beam targeting a point just medial and inferior to the PSIS. When the interosseous ligaments are engaged, the needle is rotated medially and advanced to bone on the sacrum. Radiographic contrast is injected to demonstrate spread medial to the SIJ line both cranially and caudally (see fig 2). If only cranial or caudal spread is demonstrated, then half the injectate is injected, and the needle is then directed either cranially or caudally to achieve spread in the other direction. Care is taken to avoid imbedding the tip of the needle on the ileum, which has a bulbous enlargement in this location that can serve as an alternate surface over which medication can spread (fig 3). Care is also taken to avoid a common contrast pattern in which most of the contrast flows in channels on the surface of the sacrum and directly into one of the sacral foramina (see fig 2).

View Large Image Download to PowerPoint

Fig 3. S1-3 lateral branch block: another contrast pattern to avoid.

Patient responses to the 2 injection techniques were evaluated in the following 3 ways. First, anesthetic response was graded as positive if the patient attained a pain reduction of 75% or greater than their preinjection VAS score. All scores were recorded on a pain diary with lines to record the preinjection VAS score and 8 hourly entries for the postinjection scores. Second, at the 3-week and 3-month follow-ups, pain was assessed by percentage change from the patient's preinjection VAS pain score. The follow-up period was limited to 3 months to minimize the chance that new problems might emerge, requiring additional interventions or medication changes that would complicate the comparison between the 2 SIJ injection techniques. All but 3 patients were maintained on their previous opioid regimens, which, in most cases, consisted of a variety of extended-release opioids combined with use of a breakthrough pain medicine 2 or 3 times daily. Most patients' chronic pain medication also included 1 neuromodulator, which was usually an antidepressant or anticonvulsant. Finally, at 3 weeks and 3 months, patients' change in functional status was assessed with a tool created for the pain clinic, the AIAT. This measure is a global assessment based on a patient's self-report of improvement in a basket of ADLs including housework, mobility, leisure activities, shopping, childcare, and work (if appropriate). The patient is asked to describe the overall change in these ADLs as “worse,” “no change,” “slightly improved,” “greatly improved,” or “totally restored.” The tool is scored like the patient-specified quality of life tool, a validated tool for assessing response to interventions in patients with spinal pain.31 A score of 0 (worse), 1 (no change), 2 (slight improvement), 3 (great improvement), and 4 (restored to normal). The AIAT was not previously validated in a chronic pain population, so data were collected to determine the degree of correlation with the VAS score data.

Prior studies of VAS pain-reporting tools had shown that 15% to 20% is the minimum change in VAS score usually regarded as clinically significant.32, 33, 34 For our purposes, the more stringent criterion of 50% or more change in VAS score was regarded as the threshold for treatment success. Because the AIAT specified 5 scores, it seemed reasonable to scale the AIAT for compatibility with the VAS by looking at the following correlations: (1) a negative VAS change with an AIAT score of 0 (worse), (2) VAS change between 0% and 19.99% with an AIAT score of 1 (no change), (3) VAS change between 20% and 49.99% with an AIAT score of 2 (slightly improved), and (4) VAS change of 50% or more with an AIAT score of 3 (greatly improved).

An AIAT score of 4 was not analyzed because it occurred so rarely. The AIAT and VAS correlations were assessed by determining the sensitivity, specificity, and positive-predictive value with which each AIAT score predicted a patient's VAS score changes.

The following statistical analyses were done to compare responses to the 2 injection techniques. Percentage changes in the VAS pain scores were compared by using an independent Student t test. The 2 groups' AIAT scores were compared using the Mann-Whitney rank-sum test. In addition, a Pearson Chi-square statistic was performed to check whether (1) the rates of response to treatment differed between the 2 procedures at 3 weeks and 3 months, and (2) the rates of initial anesthetic response differed between the 2 groups. Positive response to treatment was defined as a 50% or greater improvement in VAS pain rating or a score of 3 (greatly improved) or greater on the AIAT.

Analysis for potential confounding variables was done in the following way. The frequencies for all the independent variables summarized in table 1 were assessed to determine whether there were any differences between the 2 treatment groups. Pearson Chi-square statistics were calculated to determine whether any differences between the 2 groups were significant. In addition, 2 × 2 contingency tables were constructed to evaluate whether any of the variables in table 1 predicted a positive response to either sacroiliac intra-articular injection or the new modified procedure. Because the patients in the study were on a variety of opioid regimens, dosage comparisons were done by converting everyone's total daily dose to its oxycodone equivalent. To test for correlations with high-dose opiate use, patients were stratified into 3 groups in terms of opiate use: (1) use greater than or less than 120mg oxycodone equivalent a day, (2) use greater than or less than 60mg oxycodone equivalent a day, or (3) use greater than or less than 30mg oxycodone equivalent a day. Correlations were tested using the Pearson Chi-square statistic and the ϕ coefficient of association. When 2 × 2 contingency tables contained entries with very low frequencies (such as the number of diabetics in the study or the number of patients who denied symptom provocation by standing or walking), the Fisher exact probability test was performed. An independent Student t test was used to check whether there was any difference between the treatment groups' ages and initial pain scores. The Student t test could not be used to do the same cross-check analysis on symptom durations because these values were not normally distributed (see SD data, table 1). A Mann-Whitney rank-sum test was used instead. Finally, we calculated the sensitivity and specificity of the anesthetic block for predicting positive response to treatment at 3 months.

Results 

When comparing sacroiliac injection to the newer combined procedure (combined sacroiliac injection and S1–3 lateral branch block at the posterior sacroiliac ligaments), there were statistically significant differences between the changes in VAS pain scores at the 3-week and 3-month follow-ups (table 2). Similarly, the AIAT scores differed at 3 weeks and 3 months, with the differences more statistically significant at the 3-month follow-up than at the 3-week follow-up (Mann-Whitney, P=.006, P=.050, respectively). The actual distribution of AIAT scores is shown in figure 4. Furthermore, treatment response rates (defined as a 50% or better improvement in VAS score or a score of 3 or higher on the AIAT) differed significantly at 3 weeks and 3 months with the difference more significant at 3 months than at 3 weeks (table 3). In all cases, the combined procedure was superior to the standard procedure.

Table 2. Percentage Change in Pain VAS Scores at 3 Weeks and 3 Months

	Injection	Measurement Interval	Mean Initial VAS	Initial 95% CI	Mean VAS Δ	95% CI (%)	P
	Sacroiliac injection	3wk	7.88	0.35	26.3%	8.9	NA
	SI+S1-3	3wk	7.95	0.28	38.5%	7.0	NA
	Sacroiliac vs SI+S1-3						.02
	Sacroiliac injection	3mo	7.88	0.35	14.1%	6.4	NA
	SI+S1-3	3mo	7.95	0.28	29.6%	6.3	NA
	Sacroiliac vs SI+S1-3						.001

Abbreviations: CI, confidence interval; NA, not applicable; SI+S1-3, combined sacroiliac and S1-3 injection.

View Large Image Download to PowerPoint

Fig 4. AIAT score distributions at (A) 3-week follow-up and (B) 3-month follow-up. Gray bars indicate SI joint injection; black bars indicate SI + S1-S3 injection. Mann-Whitney P values at 3-week follow-up, P=.05; and at 3-month follow-up, P=.006.

Table 3. Positive Treatment Response Rate at 3 Weeks and 3 Months

	Injection	Measurement Interval	Positive Response (%)	P
	Sacroiliac injection	3wk	27.5	NA
	SI+S1-3	3wk	42.5	NA
	Sacroiliac vs SI+S1-3			.05
	Sacroiliac injection	3mo	12.5	NA
	SI+S1-3	3mo	31.25	NA
	Sacroiliac vs SI+S1-3			.025

Abbreviations: NA, not applicable; SI+S1-3, combined sacroiliac and S1-3 injection.

In addition, the 2 procedures differed with respect to the reported rates of positive response to anesthetic block (table 4). Despite the fact that the combined procedure showed a higher anesthetic block rate, the report of a positive response to anesthetic block predicted treatment response at 3 months with poor sensitivity and specificity. Conducting this analysis on the whole study population yielded a sensitivity of 88%, a specificity of 49%, and a positive-predictive value of only 44%. If we restricted that analysis to a group that was more likely to respond to treatment (eg, nonsmokers with symptom duration <24mo), the recalculated sensitivity was 90%, the specificity was 70%, and positive-predictive value was 75%.

Table 4. Differences in Response to Anesthetic Block

	Injection	Positive Response (%)	P
	Sacroiliac injection	42.5	NA
	SI+S1-3	62.5	NA
	Sacroiliac vs SI+S1-3		.037

Abbreviations: NA, not applicable; SI+S1-3, combined sacroiliac and S1-3 injection.

Comparing the 2 treatment groups, they were similar with respect to all variables studied except symptom provocation by sit-to-stand transfers. The frequency of this variable was significantly higher in the sacroiliac injection group (77.5% vs 42.5%; P=.002), but this variable did not correlate with treatment outcome. Furthermore, Student t testing showed no significant differences between the 2 treatment groups' ages or initial pain scores. Mann-Whitney analysis of the duration of symptoms reported by the 2 groups also showed no significant difference between the 2 populations.

The only epidemiologic variable that showed a significant positive correlation (P<.05) with response to treatment by the combined sacroiliac and S1-3 injection was symptom duration 24 months or less. The following variables were associated with a decreased chance of a positive outcome: symptom duration of more than 36 months, smoking, symptom provocation by walking, symptom provocation by stair climbing, and symptom provocation by standing. The strengths of these correlations are summarized in table 5.

Table 5. Variables Associated With a Response Combined Sacroiliac and S1-3 Injection

	Variable	Odds Ratio	ϕ Coefficient	P
	Symptom <24mo	3.37	.27	.015
	Symptom >36mo	0.30	−.41	<.001
	Smoking	0.28	−.27	.020
	Pain provoked by standing	0.18	−.27	.023
	Pain provoked by walking	0.21	−.29	.014
	Pain provoked by stair climbing	0.37	−.22	.046

When the same analysis was done to assess variables affecting response to SIJ injection alone, the following results were obtained. Age and duration of symptoms had a weak effect on response to treatment at 3 months. The prevalence of both of these variables was the same in our 2 treatment groups.

Finally, the AIAT and VAS correlations were based on a total of 240 data pairs, 3-week and 3-month data on 40 sacroiliac injections and 80 combined sacroiliac and S1-3 injections. The ability of the individual AIAT scores to predict the VAS changes with which they were paired is summarized in the sensitivity and specificity and positive predictive value data of table 6. With 2 exceptions, the correlations were quite strong. The sensitivity of the AIAT score of 0 (clinically worse) is low, and the positive predictive value of an AIAT score of 2 (slightly improved) is only fair.

Table 6. Correlations of AIAT Scores With VAS Changes

	AIAT Prediction of VAS Score	VAS Δ <0	VAS Δ 0%–19.9%	VAS Δ 20%–49.9%	VAS Δ >50%
		AIAT Score 0	AIAT Score 1	AIAT Score 2	AIAT Score 3
	Sensitivity	46.1	96.6	84.3	79.5
	Specificity	99.6	92.1	88.4	95.8
	Positive predictive value	85.7	87.8	66.2	89.2

NOTE. Values are percentages.

Discussion 

This study suggests that directing the corticosteroid dose not only into the SIJ but also at the posterior interosseous ligament and S1-3 lateral branches improves clinical outcome for patients with chronic sacroiliac region pain. A survey of the epidemiologic data shows the 2 treatment groups to be well matched (see table 1) despite this being a retrospective study. It should be stressed that the treatment response rates and variables associated with treatment response cited in this study may be very specific to this particular patient population, patients at a tertiary referral center for the treatment of chronic pain. Some of this study's findings have been replicated in previous studies. For instance, the finding that treatment response correlates inversely with symptom duration has been observed in many studies on spinal pain.35, 36, 37 Similarly, numerous studies have demonstrated a negative effect of smoking on the outcomes of spinal procedures, although the absence of an effect has also been observed.37, 38, 39 Some previous studies have also observed the absence of a correlation between chronic opioid use and the outcomes of spinal procedures, whereas others have shown that opioid use has a negative effect on prognosis.35, 38, 39 This study's finding that symptom exacerbation by walking, standing, and stair climbing correlated inversely with response to treatment is statistically accurate but probably is not a relevant finding. It most likely can be explained by the low treatment response rate (31.25%) combined with the extremely high prevalence of symptom exacerbation by these 3 activities. There is little scientific literature describing the effects of lumbar stenosis or prior lumbar fusion on the outcomes of sacroiliac injections. However, we were surprised by our finding that neither of these conditions had an effect on outcome, despite the fact that they often cause substantial pathology of the lumbosacral nerve roots.

The suggestion that there are important extra-articular sources of sacroiliac region pain is consistent with a number of previous studies. There are studies that demonstrate injection around the posterior interosseous ligament results in rates of improvement comparable to intra-articular injection.40, 41 Also, a recent study by Dreyfuss et al8 showed that the Fortin sign (where patients identify the PSIS and posterior ligaments as the primary site of their pain) was extremely sensitive for predicting the results of an intra-articular anesthetic block. If the primary pathology were entirely intra-articular, it is not clear why the region of the posterior interosseous ligament would be the primary focus of most patients' attention. Notably, the study by Dreyfuss8 also showed no ability to correlate numerous physical examination tests and historical features with response to diagnostic intra-articular blocks.8 Similar results have been reported by other authors, although these analyses have been contradicted in a few other published studies.42, 43, 44 If the pathology were simply intra-articular, one would expect to be able to identify some physical tests or some feature of the patients' history that demonstrated reasonably high predictive value. However, a more complex pathologic mechanism involving multiple structures, susceptible to a diverse set of provocations, might defy such clear correlations. Finally, it is a common experience in clinical practice to see sacroiliac region pain develop after painful injuries to the low back or after spinal fusions.13, 45 A number of biomechanic studies suggest that tension on various parts of the posterior sacroiliac ligaments is influenced by the actions of multiple muscles, whose activity changes in response to spinal and perispinal pain.17, 18, 19, 20, 21, 22 With recent anatomic studies showing an intimate relationship between the posterior sacroiliac ligaments and the S1-3 lateral branch nerves, changes in tension on the ligaments may have an effect on the activity of these branches.23, 24, 25, 26

These data also suggest that the literature on sacroiliac pain may be underestimating the true prevalence of pain generated from the sacroiliac region by basing those estimates purely on intra-articular diagnostic blocks. In this study, adding the extra-articular block to the intra-articular block was associated with a 47% increase in patients' reports that their presumed sacroiliac pain responded to anesthetic blockade.

Conclusions 

The data presented in this study suggest that the clinical entity we describe as chronic sacroiliac mediated pain has both intra-articular and extra-articular sources. The level of evidence presented in this article is not sufficient to claim that the combined sacroiliac and S1-3 injection technique is clearly a superior treatment technique to sacroiliac injection alone. A prospective randomized study would be required to support that claim. To assess the relative contributions of intra-articular and extra-articular sources, a 3-limb study would be required comparing intra-articular injection to both extra-articular injection alone and the combined procedure. Addition of a fourth group, placebo injection, would increase the power of such a study. This study also suggests the need for new studies on the prevalence of sacroiliac region pain. Because all the best epidemiologic data we have derive the prevalence of sacroiliac mediated pain from response to intra-articular anesthetic blocks, the literature probably underestimates the true prevalence of pain generated from the sacroiliac region. Additional work will need to be done in order to assess which sacroiliac block techniques best estimate the prevalence of sacroiliac region pain. Finally, it should be stressed that the best treatment response rate we could achieve at 3 months with corticosteroid injections was 31.25%, a rate not markedly different from the placebo response rates quoted in many studies.55, 56, 57 New, better strategies are required for the treatment of chronic sacroiliac region pain.