Outcome of Pregnancy-Related Lumbopelvic Pain Treated According to a Diagnosis-Based Decision Rule: A Prospective Observational Cohort Study
Donald R. Murphy, DC, Eric L. Hurwitz, DC, PhD, Ericka E. McGovern, DC
Rhode Island Spine Center,
Pawtucket, RI 02860, USA.
OBJECTIVE: The purpose of this study was to describe the clinical outcomes of patients with pregnancy-related lumbopelvic pain (PRLP) treated according to a diagnosis-based clinical decision rule.
METHODS: This was a prospective observational cohort of consecutive patients with PRLP. Data on 115 patients were collected at baseline and on 78 patients at the end of the active treatment. Disability was measured using the Bournemouth Disability Questionnaire (BDQ). Pain intensity was measured using the Numerical Rating Scale for pain (NRS). Patients were also asked to self-rate their improvement. Care was provided by a chiropractic physician/physical therapist team.
RESULTS: Fifty-seven patients (73%) reported their improvement as either “excellent” or “good.” The mean patient-rated improvement was 61.5%. The mean improvement in BDQ was 17.8 points. The mean percentage of improvement in BDQ was 39% and the median was 48%. Mean improvement in pain was 2.9 points. Fifty-one percent of the patients had experienced clinically significant improvement in disability and 67% patients had experienced clinically significant improvement in pain. Patients were seen an average 6.8 visits. Follow-up data for an average of 11 months after the end of treatment were collected on 61 patients. Upon follow-up, 85.5% of patients rated their improvement as either “excellent” or “good.” The mean patient-rated improvement was 83.2%. The mean improvement in BDQ was 28.1 points. The mean percentage of improvement in BDQ was 68% and the median was 87.5%. Mean improvement in pain was 3.5 points. Seventy-three percent of the patients had experienced clinically significant improvement in disability and 82% patients had experienced clinically significant improvement in pain.
CONCLUSIONS: The management strategy used in this study appeared to yield favorable outcomes in this patient population and appears to be a safe option for patients with PRLP, although because of this study’s sample size, rare complications are not likely to be detected. In addition, the absence of randomization and a control group limits interpretation with regard to clinical effectiveness. Randomized, controlled trials are necessary to distinguish treatment effects from the natural history of PRLP.
From the FULL TEXT Article
Pregnancy-related lumbopelvic pain (PRLP) is common. It has been estimated that approximately 48% to 56% of pregnant women develop lumbar and/or pelvic pain sometime during pregnancy, [1, 2] with some estimates being as high as two thirds. [3, 4] In many patients, the problem can be disabling.  In addition, women who have PRLP during pregnancy are more likely to have pain in this area during delivery. 
Noren et al  separated patients with PRLP into 3 groups, those with lumbar pain (LP), those with posterior pelvic pain (PPP), and those with a combination of both. These symptomatic groups have been found to have distinct characteristics, those with PPP having more severe functional deficit that those with LP and those with a combination of both having greater disability than either of the other groups.  Also, greater duration of sick leave has been found in patients with PPP than with LP. 
The purpose of this study is to report the outcomes of a management strategy that was founded on a diagnosis-based clinical decision rule (DBCDR), in which treatment decisions are determined by a diagnostic process that considers differential diagnostic factors, pain-generating tissues and perpetuating factors.  Outcomes of this approach have been reported in observational cohort studies in other patient populations, [9-11] but, as of yet, the approach has not been evaluated in the unique population of pregnant patients with lumbopelvic pain.
A recent Cochrane review  found 8 randomized, controlled trials of interventions for patients with PRLP. They found that stretching, strengthening, stabilization, and pelvic tilt exercises as well as acupuncture were all more effective than prenatal care alone in reducing pain.  However, the effects sizes were small for each of these treatments. It is not known whether a treatment approach that is individualized to the patient is more effective than an approach in which all patients are treated in the same manner irrespective of individual diagnosis.
In the study reported here, a strict DBCDR was applied to detect individual features in each case from which treatment decisions were made. By the end of treatment, a mean of 61.5% self-reported improvement was found and 73% of patients rated their improvement as “excellent” or “good.” Mean improvement in pain intensity was 3 points. This exceeds the 2 points determined by Farrar et al  to be the threshold for clinically significant improvement using the NRS. Clinically significant improvement in pain was found 63.5% of the patients. Improvement in disability (BDQ 39%) in the cohort as a whole did not quite reach the threshold for clinical significance of 47% ; however, the median improvement in disability was 48%, which did exceed the threshold for clinical significance.  More than half the patients experienced clinically significant improvement in disability. At long-term follow-up, however, improvements were greater. The mean self-rated improvement was 83.2%, and 85.5% of patients rated their improvement as either “good” or “excellent.” The mean percentage of improvement in disability was 68% and the median was 87.5%. These values exceed the 47% change in BDQ that Hurst and Bolton  found carried the greatest sensitivity and specificity for clinically significant change using this instrument. Mean improvement in pain at long-term follow-up was 3.5 points. This exceeds the 2 points determined by Farrar et al  to be the threshold for clinically significant improvement using the NRS.
It is interesting that improvements were greater at follow-up a mean 11 months after treatment stopped than they were at the end of treatment. Fear beliefs were also decreased at long-term follow-up as compared to at the end of treatment. This is consistent with other cohort studies of patients treated according to the DBCDR with lumbar spinal stenosis,  cervical radiculopathy,  and lumbar radiculopathy secondary to herniated disk.  This suggests that ongoing care after initial improvement in pain and disability is not typically necessary, at least when patients with these conditions are treated according to the DBCDR.
The approach taken in these studies places great emphasis on exercise, education to decrease catastrophizing and fear beliefs, and empowering patients to self-manage the condition.  This may be the reason for further improvement after release from care. However, compliance with exercise and self-management strategies was not specifically measured. Without a control group, attribution of the further improvement at long-term follow-up to the treatment cannot be made with certainty because natural history must be considered. The long-term prognosis of back pain that begins in pregnancy is similar to that of the nonpregnant population.
Although the cohort in this study reported a fairly high percentage of improvement and verbal improvement and had clinically significant improvement in pain by the end of treatment, the improvement in disability did not reach the established threshold for clinical significance. One possible explanation for this discrepancy is that some patients’ answers to the follow-up BDQ, all of which were filled out later in the pregnancy than were the initial questionnaires, may have reflected disability related to factors in the pregnancy other than the low back pain itself. The BDQ clearly requests the individual answer the questions as they relate to the back pain only; however, this was not reinforced at any time during data collection.
The location of the pain appeared to affect outcomes over the short term, but to a lesser degree over the long term. Patients who had both LP and PPP had less favorable outcomes than those who only had pain in one area. This is consistent with other studies. Ostgaard et al  found that pregnant patients with PPP faired more poorly to individual and group back school classes than did patients with LP. They did not include a group that had pain in both locations. Gutke et al  found that patients with both LP and PPP had greater pain intensity and disability and lower health status index and self-rated health status than those with pain in only one area. Noren et al  also found greater disability in patients with the combination of LP and PPP than in those with only one pain location. Patients who had LBP in a previous pregnancy generally faired poorer than those who did not over the short term and the long term. This is consistent with previous studies that have found this historical finding as a risk factor for the development of PRLP. [1, 47] Interestingly, previous history of LBP was not found in this study to affect outcome. However, with regard to all the factors that were assessed for influence on outcome, a larger sample size would be required to draw firm conclusions.
Transient increase in pain after a single session was seen in 3.4% of patients. No major complications were seen in any patient. The rate of transient increase in pain is substantially lower than other studies of treatments that have focused on manual therapy in which the rate was approximately 1/3. [48, 49] Although 10 of the 78 subjects for whom complete data were obtained were not treated with manual therapy, this still would suggest that the treatment approach taken here is safe for this patient population. This conclusion is not limited by the uncontrolled nature of the design, as observational designs are considered adequate for drawing conclusions about safety.  However, a larger sample size would be needed to detect rare complications.
The findings of this study are interesting in light of previous research. Lisi  reported a retrospective case series of 17 patients treated with an eclectic approach similar to that used in the current study. He found a mean improvement in NRS of 4.4 points. Disability and other outcomes were not measured. The retrospective design and small sample size do not allow direct comparisons with the current study.
There was significant loss to follow-up in this cohort. Baseline data were gathered on 115 subjects, but data at first reexamination were only obtained in 83 subjects, and long-term follow-up data were obtained in 61 subjects. This is substantially greater than the loss to follow-up in the 3 other cohort studies of the DBCDR performed by the authors of patients with lumbar spinal stenosis,  cervical radiculopathy,  and lumbar radiculopathy secondary to herniated disk.  All these studies were practice-based cohort studies and, thus, it was not possible to use the strict controls that are typical of a funded clinical trial. However, one factor in the present study that made it different from the other cohort studies was that this was a cohort of pregnant patients with lumbopelvic pain. As such, a number of patients delivered before any reexamination data could be gathered. Some others developed pregnancy-related problems apart from the lumbopelvic pain that required bed rest or specialized medical intervention. In addition, there were a number of patients with Medicaid, which does not cover chiropractic care in Rhode Island. A number of these patients attended the first visit but did not return. Attempts were made to accommodate these patients but, again, as this was an unfunded study, this was difficult. Of the patients who were reachable by phone, 3 stated that they did not return because they were feeling better after the initial treatment sessions. No patient who was reached stated that they did not return because they felt worse or did not improve.
This is an observational cohort study; thus, the absence of randomization limits interpretation of the findings regarding outcome. A no-treatment comparison group would help to distinguish specific treatment effects from the natural history of PRLP. In addition, because an eclectic approach was used, there is no way to determine the extent to which any particular treatment influenced the outcomes. On the other hand, as the study was conducted in a “real-world” setting, practitioners who have the appropriate training can apply the protocol in the office setting. Also, it allows for the assessment of an approach that individualized the treatment for each patient.
Reported here are short- and long-term outcomes of treatment founded on a DBCDR in patients with PRLP. It appears that this approach may be beneficial for a substantial proportion of these patients and that further improvement occurs over time, even after cessation of treatment. However, firm conclusions cannot be drawn the absence of a control group. It appears that the treatment approaches used in this study are safe for this patient population; however, a larger sample size would be necessary to detect rare complications. Further investigation in the form of randomized, controlled trials is called for to determine the efficacy of this approach.