Side Effects and Adverse Events Related to Intraligamentous Injection of Sclerosing Solutions (Prolotherapy) for Back and Neck Pain: A Survey of Practitioners

Methods 

Results 

Training and Experience 

The most common degree reported was doctor of medicine (MD) (n=104), followed by doctor of osteopathic medicine (DO) (n=60). The most common medical specialties were physical medicine (n=53) and general practice (n=49). One hundred thirty-five of the 153 respondents were board certified in related disciplines. Respondents had learned prolotherapy through continuing medical education courses and by observing colleagues. Respondents had a median of 10 of years of experience with prolotherapy for back and neck pain, during which time they treated a median of 500 patients and provided a median of 2000 treatments. The spinal region treated most often with prolotherapy was lumbosacral (median, 60%) (Table 1, Table 2).

Table 1. Training of Practitioners Performing Prolotherapy for Back and Neck Pain

	Variable	n
	Degree
	Doctor of medicine	104
	Doctor of osteopathy	60
	Doctor of naturopathy	2
	Nurse practitioner	2
	Medical specialty
	Physical medicine	53
	General practice	49
	Other	21
	Anesthesiology	18
	Orthopedics	14
	Internal medicine	10
	Board certified
	Yes	135
	No	28
	Board discipline
	General practice	36
	Pain management	25
	Other	21
	Physical medicine	20
	Anesthesiology	2
	Prolotherapy training
	AAOM course	99
	Other	85
	Observing a colleague	80
	Ravin course	66
	Hackett course	60

Table 2. Experience of Practitioners Offering Prolotherapy for Back and Neck Pain

	Variable	Median	Mean ± SD	Range	Lower Quartile⁎	Upper Quartile†
	Years of experience
	In practice	21	22±12	1–63	14	30
	With spinal prolotherapy	10	12±9	1–63	6	15
	Total no. of spinal prolotherapy
	Patients	500	2279±9234	1–110,000	100	1625
	Treatments	2000	11,042±52,595	3–660,000	400	6000
	% of spinal prolotherapy treatments in the
	Cervical region	25	23±15	0–100	10	20
	Thoracic region	10	12±9	0–45	5	20
	Lumbosacral region	60	65±10	0–100	50	80

Abbreviation: SD, standard deviation.

⁎ Lower quartile is the 25th percentile. † Upper quartile is the 75th percentile.

Side Effects 

The side effect with the highest estimated median prevalence was pain (70%), followed by stiffness (25%), bruising (5%), and temporary numbness (1%). Other side effects listed on the questionnaire (ie, headache, nausea and vomiting, skin infection, temporary numbness, temporary weakness) had a median reported prevalence of 0% (table 3).

Table 3. Estimated Prevalence (%) of Side Effects Associated With Prolotherapy for Back and Neck Pain

	Variable	Median	Mean ± SD	Range	Lower Quartile⁎	Upper Quartile†
	Bruising	5	13.9±20.7	0–100	1	20
	Headache	0	1.8±7.9	0–100	0	1
	Nausea and vomiting	0	0.8±1.8	0–10	0	1
	Pain	70	57.9±36.7	0–100	21	95
	Skin infection	0	0.2±1.9	0–25	0	0
	Stiffness	25	36.3±36.5	0–100	0	75
	Temporary numbness	1	13.4±25.5	0–100	0	10
	Temporary weakness	0	1.4±3.8	0–30	0	1

⁎ Lower quartile is the 25th percentile. † Upper quartile is the 75th percentile.

Adverse Events 

There were 472 adverse events related to prolotherapy for back and neck pain reported in this survey. The most commonly reported adverse events were cases of spinal headache (n=164), pneumothorax (n=123), temporary systemic reactions (n=73), and nerve damage (n=54). A total of 69 adverse events required hospitalization with the most common reason being pneumothorax (n=48). A total of 5 adverse events resulted in permanent injury, all secondary to nerve damage (table 4).

Table 4. Number of Adverse Events Associated With Prolotherapy for Back and Neck Pain

	Variable	Total	Requiring Hospitalization	Resulting in Permanent Injury
	Diabetic complications	2	0	0
	Spinal disk injury	2	1	0
	Hemorrhage	27	2	0
	Infection	18	4	0
	Nerve damage	54	1	5
	Pneumothorax	123	48	0
	Spinal headache	164	6	0
	Spinal cord insult	9	0	0
	Systemic reactions	73	7	0
	Total	472	69	5

After compiling data from respondents, we made 2 revisions to the adverse event categories. The adverse-event categories meningitis/paralysis/spinal cord injury, presented separately in the survey, were combined into “spinal cord insult.” This revision was made after reading comments and detailed adverse-event reports indicating that respondents were using these categories interchangeably to describe temporary symptoms such as lost or diminished sensation or strength in the lower limbs and so on. Similarly, the adverse event categories anaphylaxis/cardiopulmonary events/shock/systemic toxicity were combined into “temporary systemic reactions.” This revision was made after reading detailed adverse-event reports and comments indicating that respondents were using some of these categories interchangeably to describe various types of rapid-onset, short-lived symptoms such as presyncope, syncope, dizziness, hypotension, and bradycardia. The revised categories better reflect the nature of the adverse events reported.

Discussion 

To our knowledge, data from this survey provide the most comprehensive understanding to date of the training and experience of practitioners offering prolotherapy for back and neck pain, their use of specific treatment procedures, the estimated prevalence of side effects, and the nature and sequelae of adverse events related to this treatment approach. With regard to training, almost all (98%) respondents held MD or DO degrees, and many (83%) were board certified in related disciplines such as physical medicine and pain management. This indicates that prolotherapy, commonly considered a CAM treatment, is being practiced by conventionally trained physicians. In fact, some respondents reported learning about prolotherapy through conventional medical education such as a rotation (n=1), residency (n=2), fellowship (n=2), or preceptorship (n=2). Respondents were well-experienced clinicians with a median of 21 years in practice and 10 years administering prolotherapy for back and neck pain. The finding that the respondents provided a median of 2000 prolotherapy treatments to 500 patients agrees with published literature indicating that patients often receive 4 to 6 treatments at weekly to monthly intervals during the course of care.2 The total number of patients treated by all respondents numbered in the hundreds of thousands, which is suggestive of a strong demand from patients for prolotherapy.

A clear majority of respondents reported injecting drug solutions that contained dextrose, glycerin, phenol, and lidocaine, whereas less than 2% still used zinc, one of the inappropriately strong chemical irritants that was implicated in serious adverse events related to prolotherapy in the 1950s.6 Although reasons for favoring particular drug ingredients in prolotherapy solutions were not sought from respondents, conversations with practitioners have indicated that, absent compelling efficacy data regarding specific ingredients, those with greater perceived safety (eg, dextrose) are generally favored. Most practitioners obtained these drug solutions from pharmacies, although a sizeable portion reported preparing these themselves, making sterility a potential concern. However, because admixtures for injection (eg, lidocaine, methylprednisolone) are commonly prepared in physician offices without difficulty, it is reasonable to assume that customary vigilance regarding aseptic preparation is also being observed for prolotherapy solutions. Almost all (96%) respondents used common skin sterilization before the injections, thus minimizing the risk of infection.

A majority (87%) of respondents did not administer intravenous sedation before the injection procedure, thus reducing a potential source of complications. The heterogeneity in the use of specific treatment procedures such as the composition and volume of the drug solution injected, for example, may be attributable to a lack of standardization in the treatment approach taught in various continuing medical education courses on the subject.

The nature and estimated prevalence of side effects reported in this survey are consistent with observations from a randomized controlled trial of prolotherapy for chronic low back pain, which reported the following symptoms in its participants at least once during the course of care: increased pain (88%), increased stiffness (76%), headache (59%), nausea and diarrhea (42%), and other symptoms (56%).7 Discrepancies in these figures and those reported in our survey are likely caused by differences in study methodology. Common side effects such as temporary pain and stiffness are consistent with a commonly proposed mechanism of action for prolotherapy. This theory purports that injection of irritants produces a controlled acute inflammatory response to stimulate fibroblastic proliferation and connective tissue repair.3 Although these side effects appear benign and short lived, they may present a relative contraindication to this treatment for patients with a low pain threshold who could be adversely impacted by a temporary increase in pain. If practitioner are able to identify such patients a priori, clinical experience suggests that certain modifications (eg, light sedation, injection of smaller volumes, or lower than usual concentration of irritants) may be appropriate to minimize postinjection pain.

The majority (61%) of adverse events reported in this study consisted of spinal headache and pneumothorax. Both of these adverse events were easily treated, and neither resulted in any permanent sequelae. Other adverse events reported in this survey that may also have been caused by injection technique include disk injuries, nerve damage, spinal cord injuries, and hemorrhage. These injuries are similar in nature to those routinely reported after other common spinal injection procedures such as facet joint, epidural, or local injections.8, 9, 10, 11, 12 Many spine clinicians perceive these common injection procedures as safe and prescribe them for their patients on a regular basis, despite lack of compelling evidence for their long-term efficacy.12 It is also important to note that these injuries can occur irrespective of the drug being injected. Many of the detailed adverse event reports in our survey indicate that needle injuries occurred when probing for the desired injection site with the needle tip, before delivering the intended bolus of drug. Although we are aware that fluoroscopy is occasionally used for needle guidance by certain prolotherapy practitioners, this equipment is not widely available to many practitioners. Moreover, there is currently no evidence that this procedure leads to fewer needle injuries.

The other major group of adverse events reported in this survey consisted of temporary systemic reactions. Several comments from respondents indicated that these reactions were likely attributable to the local anesthetics contained in the solution, some of which are known to have the potential for these effects.13, 14, 15 We were unable to ascertain if any true cases of systemic toxicity confirmed by abnormal laboratory values may have occurred as a result of adverse reactions to various components of the solution injected. The absence of reported systemic toxicity may indicate that practitioners do not routinely monitor their patients for this possibility. This finding, however, is consistent with a previous randomized clinical trial that found no difference in pre- and posttreatment laboratory testing (ie, complete blood count, sedimentation rate, chemistry panel, thyroid function, urinalysis) between the group assigned to dextrose/glycerin/phenol injections or saline injections.16

The nature of adverse events reported in this study was similar to those from a previous survey of prolotherapy practitioners by Dorman17 in 1993. The 95 respondents in that study had collectively treated 494,845 patients for a variety of conditions, with the most common indication being low back (69%) followed by other areas of the spine (20%). Respondents to that survey reported a total of 65 “complications.” The most commonly reported complication was 29 cases of pneumothorax (2 of which required chest tubes), followed by 24 cases of allergic reactions and 12 other cases requiring hospitalization or resulting in transient or permanent nerve damage; few details were provided regarding the complications or their sequelae. That survey also reported that most respondents used a combination of dextrose, glycerin, and phenol in their prolotherapy drug solutions, which concurs with the results of our study.

We are well aware of the limitations and biases inherent to this study design. One of the most likely limitations of the study is recall bias. Given the design of the study (ie, written survey), recall bias may have played a role in the responses. Future researchers should account for potential recall bias in their study design. Considering the response rate of approximately 50%, 1 limitation is a potential participant bias. Given the number of participants who reported and provided details on adverse events, as well as the anonymity provided by our methods, the effect of participant bias was likely minimized. Another limitation is that physicians practicing prolotherapy for back and neck pain who were not members of the AAOM or ACOPMS were not surveyed; differences in the outcomes measured that may exist between members and nonmembers are unknown. There is also possible estimation bias, whereby large numbers such as total treatments may be overstated and small numbers such as adverse events may be understated. We thus declined to provide potentially misleading ratio data. Similarly, we refrained from making direct statements regarding the rate of reported adverse events per injection in this study given the uncertainty and high potential for recall bias in estimating both the numerator (ie, rare and likely underreported adverse events over a long period) as well as the denominator (ie, total number of treatments given and patients treated). Another reason for not emphasizing rates per injection or treatment was that the data reported in this survey were not normally distributed and skewed heavily by outliers, which would have made any reported rates difficult to interpret.

Nevertheless, data on rare adverse events are seldom captured in small clinical trials. Any insight about their nature and estimated occurrence gathered from physician surveys capturing a large patient population is useful to plan additional studies on the topic and provides interim data to both physicians and patients. Despite the limitations of data obtained from practitioner surveys, this study represents the most thorough examination of the nature, number, and severity of adverse events and side effects related to prolotherapy for back and neck pain and overcomes the low statistical power of clinical efficacy studies in estimating the incidence of rare adverse events.

Conclusions 

The most frequently reported adverse events related to prolotherapy are similar to those associated with other commonly used spinal injection treatments and included spinal headache, pneumothorax, and nerve damage. Common side effects included pain, stiffness, and bruising, consistent with the injection procedure and purported mechanism of action of acute inflammation. The majority of practitioners offering prolotherapy for back and neck pain are board-certified physicians who learned this treatment approach through continuing education courses. Although heterogeneity exists in the use of specific treatment procedures, the use of drug solutions containing ingredients previously associated with serious adverse events has been discontinued. To better assess the true risk of adverse events related to prolotherapy for back and neck pain, future studies designed to overcome biases inherent to practitioner surveys should be conducted. Such studies should include (1) confirmation of self-reported survey data through independent audit of patient records for adverse events and (2) prospective, multicenter, longitudinal cohort studies of patients receiving prolotherapy to monitor and record adverse events. Data from these additional studies could then be used to more accurately estimate and evaluate any risks associated with this treatment approach.