Objective
The use of epidural steroid injections has increased dramatically, but knowledge of potential adverse effects is lacking. An association between steroid injection and subsequent abnormal vaginal bleeding has been suspected clinically, but evidence has been limited to anecdotal reports.
Study Design
Paired observational retrospective cohort study using electronic medical records from a large integrated health care system. Participants were all nonhysterectomized women who underwent epidural steroid injections in 2011. For each steroid injection, encounters for abnormal vaginal bleeding during the 60 days preceding and 60 days after the injection were compared as paired observations. For women found to have bleeding, medical records review was performed to examine menopausal status and bleeding evaluation outcomes.
Results
Among 8166 epidural steroid injection procedures performed on 6926 nonhysterectomized women, 201 (2.5%) procedures were followed by at least 1 outpatient visit for abnormal vaginal bleeding. Women were 2.8 times more likely to present with abnormal vaginal bleeding during the postinjection period compared with the preinjection period ( P < .0001). Of the 197 women with postinjection bleeding, 137 (70%) were premenopausal and 60 (30%) were postmenopausal. Postinjection bleeding prompted endometrial biopsy evaluation in 103 (52%) cases, with benign findings for 100% of premenopausal women (59/59) and 95% of postmenopausal women (42/44).
Conclusion
Epidural steroid injections are associated with subsequent abnormal vaginal bleeding for both premenopausal and postmenopausal women. Women undergoing epidural steroid injection should be advised of abnormal bleeding as a potential adverse effect and providers should be aware of this association when evaluating abnormal bleeding.
Epidural steroid injection is a common treatment for lumbar and cervical radiculopathy. In the United States, the number of Medicare patients receiving such injections grew by 121% percent between 1997 and 2006 at an annual rate of 13.5%, with approximately 2.5 million epidural procedures performed in 2011. The number of procedures among non-Medicare patients is not easily assessed but is thought to be of similar or greater volume. Although use of these injections is increasing dramatically, research on their potential adverse effects has been relatively limited. Known risks include infection, nerve injury, and epidural hematoma, as well as risks associated with corticosteroid use. An association between steroid injections and subsequent abnormal vaginal bleeding has never been demonstrated, although the phenomenon has been reported anecdotally. Given the absence of formal study to date, we sought to examine whether epidural steroid injections lead to abnormal vaginal bleeding episodes among a population-based cohort of women undergoing epidural steroid injection within a closed integrated health care system.
Materials and Methods
Kaiser Permanente Northern California is a comprehensive health care delivery system that provides care for over 3.5 million members and includes 19 separate medical centers, with membership representing approximately 35% of the population in the geographic areas served by the health plan. As a closed integrated system with electronic medical records, all outpatient visits for both epidural steroid injection and abnormal vaginal bleeding are captured within the same databases. After approval by the Institutional Review Board for Health Services, women who received an outpatient epidural steroid injection at any of the health system clinics during 2011 were identified using a database linked to electronic medical records, searching for the Common Procedural Terminology (CPT) codes for procedures specific to epidural steroid injection (Lumbar translaminar injection 62311, Lumbar transforaminal injection 64483/64484, Cervical translaminar injection 62310, Cervical transforaminal injection 64479). Procedures having CPT subcodes indicating injection of only anesthetic were excluded. Within the health system, epidural steroid injections are performed in outpatient clinics by either Pain Medicine or Anesthesia providers who have completed a pain medicine fellowship. All injections are performed with fluoroscopic guidance to confirm placement of steroid in the epidural space or, in the case of transforaminal injection, at the nerve root level. Electronic medical records review for validation of procedure code and confirmation of steroid dose was done for 300 procedures.
A control group for comparison of the rate of abnormal vaginal bleeding was identified from women who had an outpatient office visit with a diagnosis of low back pain ( International Classification of Diseases, Ninth Edition [ICD-9] codes 724.2, 724.3, 724.4, 724.03) in 2011, but who did not receive an epidural steroid injection during this time. From this group, women were randomly selected to match the age distribution of the cohort receiving epidural steroid injection. Age at the time of epidural steroid injection or the visit for low back pain, body mass index, and self-reported race/ethnicity were identified from health plan databases. For both groups, women with a history of hysterectomy before the study period were excluded.
A paired observational design was used. For women who received an epidural steroid injection, the preinjection period was defined as 60 days before the procedure, whereas the 60 days after the procedure defined the postinjection period. We selected 60 days based on the case reports describing abnormal vaginal bleeding up to 4 weeks after injection, and allowing for additional time following this to schedule and come in for an appointment. Similar time periods for the women who were seen for back pain but did not receive an injection were defined as before and after their outpatient visit for low back pain. For each patient, the electronic database was searched to identify outpatient encounters during both time periods, in which a diagnosis indicating abnormal vaginal bleeding (ICD-9 codes: 626, 626.2, 626.3, 626.4, 626.5, 626.6, 626.7, 626.8, 627. 627.1) was recorded. For women who had more than 1 epidural steroid injection during 2011, injections done less than 120 days after a previous injection were excluded, to avoid overlap between pre- and postinjection periods. For both groups, we compared the proportion having an outpatient encounter for abnormal vaginal bleeding during the periods before and after each injection procedure, as paired observations.
Electronic medical records were reviewed for each case of postinjection bleeding to confirm the medication and dose given and the menopausal status from physician notes. If endometrial biopsy or imaging was obtained for evaluation of bleeding, the results were recorded. Hysterectomies done for evaluation of bleeding, or for treatment of bleeding within 3 months of the encounter for postinjection bleeding, were identified and the surgical pathology results noted.
Analyses were performed using SAS software version 9.13 (SAS Institute Inc, Cary, NC) with 2-sided P value of less than .05 considered significant. We compared the proportion with an outpatient encounter for abnormal vaginal bleeding during the preinjection and postinjection periods for each epidural steroid injection using McNemar’s test. Conditional logistic regression analysis was also conducted. We expected, with a sample size of 1219 pairs, to have 80% power to detect a difference of 4% between the proportions with abnormal bleeding preinjection compared with postinjection, when using an estimate of the proportion of discordant pairs of 0.25 in McNemar’s test of equality of paired proportions with a 0.05 2-sided significance level.
Results
We identified 6926 nonhysterectomized women who underwent an epidural steroid injection procedure in 2011, with 16.8% of women having more than 1 procedure during that time for a total of 8166 epidural steroid injections. Of these injections, 4497 were done using a translaminar approach, and 3669 used a transforaminal approach. Medical records review confirmed accurate coding of these procedures with epidural steroids administered in all 300 cases reviewed. We found the dose of steroid delivered to be 80 mg of either triamcinolone acetate or methylprednisolone acetate (medications considered to be of equivalent steroid potency) in 99% of translaminar procedures, and between 40 mg and 80 mg in 98% of transforaminal procedures.
There were 81 epidural steroid injections on 79 women that were associated with an encounter for abnormal vaginal bleeding in the 60 days before injection. Bleeding encounters were seen in the 60 days after the procedure for 201 injections done on 197 women. Although the average age of women having the procedure was 61.8 years, women with bleeding were younger, with an average age of 50.7 years for women with preinjection bleeding and 47.2 years for those with postinjection bleeding. The age and race/ethnicity distribution and body mass index range for the overall cohort and for women with bleeding is shown in Table 1 .
| Characteristic | All women, n (%) | Women with preinjection bleeding a | Women with postinjection bleeding |
|---|---|---|---|
| n = 6926 | n = 79 | n = 197 | |
| Age, mean (SD) | 61.8 (15.2) | 50.7 (12.6) | 47.2 (12.0) |
| Age group | |||
| <30 | 145 (2.1) | 2 (2.5) | 10 (5.1) |
| 30-39 | 452 (6.5) | 6 (7.6) | 48 (24.4) |
| 40-49 | 891 (12.9) | 36 (45.6) | 62 (31.5) |
| 50-59 | 1462 (21.1) | 21 (26.6) | 46 (23.4) |
| 60-69 | 1670 (24.1) | 5 (6.3) | 22 (11.2) |
| 70-79 | 1401 (20.2) | 7 (8.9) | 8 (4.1) |
| >79 | 905 (13.1) | 2 (2.5) | 1 (0.5) |
| Race/ethnicity | |||
| White | 4463 (64.4) | 41 (51.9) | 102 (51.8) |
| Black | 706 (10.2) | 13 (16.5) | 35 (17.8) |
| Hispanic/Latino | 778 (11.2) | 13 (16.5) | 32 (16. 2) |
| Asian/Pacific Islander | 495 (7.2) | 6 (7.6) | 16 (8.1) |
| Mixed race, other | 425 (6.1) | 6 (7.6) | 10 (5.1) |
| Unknown | 59 (0.9) | 0 | 2 (1.0) |
| BMI | |||
| Missing | 35 (0.5) | 0 | 0 |
| ≤25 | 2390 (34.5) | 25 (31.7) | 66 (33.5) |
| 26-30 | 2049 (29.6) | 21 (26.6) | 47 (23.9) |
| >30 | 2452 (35.4) | 33 (41.8) | 84 (42.6) |
| 1 ESI procedure a | 5763 (83.2) | 77 (97.5) | 193 (98.0) |
| >1 ESI procedure | 1163 (16.8) | 2 (2.5) | 4 (2.0) |
a Procedures done <120 days after previous procedure were excluded.
For each of the 8166 epidural steroid injections, outpatient encounters for abnormal vaginal bleeding during the preinjection period and postinjection period were identified. When paired analysis was performed, we found that 15 procedures were associated with bleeding both before and after injection, leaving 186 procedures associated with postinjection bleeding exclusively. When conditional logistic regression was performed, the odds ratio for bleeding in the postinjection period compared with the preinjection period was 2.8 (95% confidence interval, 2.1–3.7, P < .0001).
When we analyzed the age-matched group of 6926 women with low back pain who did not have an epidural injection, we found the rate of abnormal vaginal bleeding both before and after the visit for low back pain to be low, with 67 women having encounters for abnormal bleeding in the 60 days before their outpatient visit, compared with 77 women seen in the 60 days after the visit. When paired analysis was performed, there was no difference in the likelihood of having abnormal bleeding in the time period after the visit compared to before ( Table 2 ).
| Variable | Bleeding before (+) after (+), n | Bleeding before (+) after (−), n | Bleeding before (−) after (+), n | Bleeding before (−) after (−), n | McNemar χ 2 test, P value | OR (95% CI) |
|---|---|---|---|---|---|---|
| 6926 women undergoing 8166 epidural steroid injections | 15 | 66 | 186 | 7899 | 57.14 < .0001 | 2.8 (2.1–3.7) |
| By age ≤ or >55 y | ||||||
| 2336 women ≤55 who had 2664 injections | 11 | 46 | 141 | 2466 | 48.26 < .0001 | 3.0 (2.2–4.2) |
| 4590 women >55 who had 5502 injections | 4 | 20 | 45 | 5433 | 9.62 .002 | 2.3 (1.3–3.8) |
| Low back pain cohort (n = 6926) women with office visit for low back pain | 10 | 57 | 67 | 6792 | 0.81 .37 | 1.2 (0.8–1.7) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
