Association between radicular low back pain and constipation: a retrospective cohort study using a real-world national database : PAIN Reports

Journal Logo


Association between radicular low back pain and constipation: a retrospective cohort study using a real-world national database

Trager, Robert Jamesa,*; Mok, Shaffer R.S.b; Schlick, Kayla J.c; Perez, Jaime A.c; Dusek, Jeffery A.a

Author Information
PAIN Reports 6(3):p e954, September/October 2021. | DOI: 10.1097/PR9.0000000000000954


Previous observational studies have reported an association between lumbosacral radiculopathy (LSR), a form of low back pain (LBP) with nerve root involvement, and constipation. However, it is unclear whether this association is due to confounding variables such as comorbidities and medications.


This study explores the possible association between LSR and constipation, with the hypothesis that adults with LSR have increased odds of developing constipation compared with those with nonradicular LBP.


Adults aged 18 to 49 years with incident LSR and nonradicular LBP were identified from a national 70 million patient electronic health records network (TriNetX). Propensity score matching (PSM) was used to control for covariates and determine the odds ratio (OR) of constipation over a 1-year follow-up. Lumbar stenosis, cauda equina syndrome, and inflammatory bowel diseases were excluded.


After PSM, 503,062 patients were in each cohort. Before PSM, the likelihood of constipation was identical between cohorts (LSR 10.8% vs 10.9%; OR [confidence interval] = 0.99 [0.98-1.0], P = 0.251). This association was unchanged after PSM (LSR 10.8% vs 11.1%; OR [confidence interval] = 0.98 [0.97-0.99]; P = 0.003).


The study hypothesis can be refuted given that the OR approximated the null in a large propensity-matched sample. Patients with LSR have equivalent odds of constipation compared with those with nonradicular LBP, suggesting that LSR is not a direct cause of constipation. The similar risk of constipation between cohorts could be explained by factors common to LBP in general, such as pain severity, physical inactivity, and constipating medications.

1. Background

Lumbosacral radiculopathy (LSR) is a subset of low back pain37 (LBP) affecting one or more lumbar or sacral nerve roots that can involve radiating pain into the lower extremity, weakness, and/or sensory loss.41 Lumbosacral radiculopathy is pathoanatomically and clinically distinct from nonradicular LBP, which has different clinical features8 and treatment.25 However, LSR is less common than nonradicular LBP, having a point prevalence of 2% to 13%.18 Degenerative conditions of the low back such as lumbar disk herniation (LDH) and lumbar spinal stenosis (LSS) are the most common causes of LSR, although rarely LSR results from nondegenerative conditions such as infections or neoplasms.41

Constipation is a common symptom that affects 3% to 27% of the adult population28 and often involves neurological mechanisms.26 It is defined as a change in bowel habits involving a reduced frequency of defecation or change in stool caliber to hard or dry feces. Subcategories of constipation include slow transit, in which transit time is increased; defecatory disorders, which involve a decreased ability to evacuate stool from the anorectum; and constipation-predominant irritable bowel syndrome (IBS), which involves pain associated with altered bowel habits.26

There is neuroanatomical evidence that the large intestine receives innervation from the lumbosacral nerve roots. However, colonic motility is self-regulated by the enteric nervous system, and it is unclear to what extent this lumbosacral innervation influences motility. The proximal colon is innervated by lumbar splanchnic nerves while the distal colon is innervated by lumbar splanchnics and sacral pelvic nerves.4 The cell bodies of these nerves are located within the lumbosacral dorsal root ganglia.4 These afferent nerves enable the perception of colonic fullness4 and are believed to mediate colonic motility through a spinal defecatory reflex that engages the parasympathetic nervous system.34

Owing to the close neuroanatomical relationship between the colon and lumbosacral nerve roots, we suspected that a neurological lesion of these nerves (ie, LSR) could reduce colonic motility. Our hypothesis is based on the possibility for LSR to (1) interrupt the spinal defecatory reflex, leading to decreased colonic parasympathetic activity and motility and (2) reduce afferent signaling from the colon, leading to reduced colonic sensation and a corresponding decreased urge to evacuate.

The ancient physician Hippocrates may have been the first to propose an association between LSR and constipation. He noted that patients with sciatica who required medicine to move their bowels would have longer-lasting symptoms.13 He and other ancient physicians recommended laxatives for these patients.23 Despite these historical anecdotes, there has been limited research exploring the association between LSR and constipation.

One small study of treatments for radicular sciatica reported that 45% of patients were constipated at baseline.1 Another study documented a high incidence of constipation in patients after thoracolumbar fusion surgery.36 A study documenting the disability level of patients with LSS noted that constipation was a common complaint, but this finding was not further investigated.50 However, because these studies have a small sample size and did not control for confounders, it is not possible to make firm conclusions about the etiology of constipation in these patients.

One major confounder in this area of research is that medications commonly used for LSR increase the risk of constipation. An example of this is from a study that found acetaminophen and nonsteroidal anti-inflammatory drugs (NSAIDs) had significant associations with constipation.5 Opioids are a well-known risk for constipation, with 40% to 95% of patients developing this side effect.22 There is some evidence that other medications for LSR7 including benzodiazepines3,32 and pregabalin15 also increase the risk of constipation.

Bowel or bladder dysfunction has been better described in patients with cauda equina syndrome (CES), a more severe type of lumbosacral nerve root injury compared with LSR.40,41 Constipation has been reported in patients recovering from surgery for discogenic CES29,40 and in those developing CES as a surgical complication.24This study sought to determine whether patients with LSR have greater odds of developing constipation compared with those with nonradicular LBP, before and after controlling for covariates using propensity score matching (PSM). We hypothesize that adults with LSR have greater odds of constipation than those with nonradicular LBP. To the best of our knowledge, no previous study has investigated an association between LSR and constipation using a large database of patients while controlling for confounding variables.

2. Materials and methods

2.1. Study design

This study followed a preregistered protocol44 available at and is reported according to the STROBE statement.48 The original protocol was modified to improve the specificity of the LSR cohort by excluding patients with LSS (by code and age restriction), omitting the ICD-9 inclusion codes for radiculitis of “unspecified” regions of the spine (ie, 729.2 and 724.4), and reduce confounding by propensity matching for IBS. This study incorporates a new-user, active-comparator, retrospective observational design using aggregated, multisite, EHR-based real-world data. The patient population includes adults from age 18 to 49 years of any sex.

2.2. Setting and data source

This study used a subset of the TriNetX (TriNetX, Inc., Cambridge, MA) research network, which has been described previously.43 This is a federated national network that includes aggregated, deidentified data from over 70 million patients across EHRs of 51 health care organizations (HCOs). Information relating to participating HCOs is kept anonymous by TriNetX; therefore, geographic or institutional information are unavailable. However, a typical participating HCO represents a large academic health center with inpatient, outpatient, and specialty care services. To safeguard protected health information, TriNetX deidentifies data by restricting the population to patients younger than 90 years in accordance with the Health Insurance Portability and Accountability Act, which considers age >90 years as an identifier. In addition, queries that return patient counts <10 are rounded up to 10. These safeguards were not pertinent to our study which focused on patients <50 years and those with constipation which is a common outcome. TriNetX enables queries using standardized terminologies such as International Classification of Disease (ICD) and Current Procedural Terminology codes.

At University Hospitals of Cleveland, access to TriNetX is managed by the University Hospitals Clinical Research Center. A TriNetX search query was performed on June 16, 2021, to identify patients admitted between January 1, 2004, and June 16, 2020. Studies of this type are determined by the University Hospitals Institutional Review Board to be exempt from requiring informed consent. An individual level data set was not needed because of the study aims and required statistical analyses. This type of data set is necessary when the statistical techniques to be used exceed the capabilities of the analytical tools available within TriNetX. For this study, the statistical tools within the TriNetX platform were sufficient to compare baseline characteristics and perform PSM, and individual patient data that would be needed to conduct additional testing such as a multivariate logistic regression for an outcome were not necessary.

2.3. Participants

A new-user, active-comparator design was used in this study to reduce prevalent user bias and increase the comparability of cohorts.12 Because TriNetX identifies patients with the first occurrence of a given diagnosis by default, our study included patients with new-onset (incident) LBP. The new-user design was applied to the outcome of interest by excluding patients with a history of constipation (Supplementary material, Figure 1, available at An active comparator of LBP was used instead of healthy controls because these patients have a related health care problem causing them to seek care, which allowed for greater comparability to the LSR cohort.

2.4. Eligibility criteria

Adult patients were included who had incident, nonurgent, radicular, or nonradicular LBP and no previous history of constipation. An age range of 18 to 49 years was used to make the LSR cohort more specific to radiculopathy resulting from LDH rather than LSS. Previous studies have shown that LSR related to LDH is more common in patients in their 30s and 40s,14,52 and becomes less common after age 50, during which point LSS becomes more common.14 A cutoff point of 50 years has been used previously to help distinguish between patients with LDH and LSS in claims data.11

Specific exclusions for both cohorts were used to make the query more specific: (1) neurological disorders at an alternate anatomic site that could cause LBP, (2) serious, urgent, or emergent low back–related pathology, and (3) inflammatory bowel disease (Supplementary material, Table 1, available at Severe low back–related pathology such as CES, neoplasms, infection, and fractures were excluded by using ICD-9 codes that were used in previous studies for this purpose.6,10,17 We chose not to exclude an exhaustive list of neurological disorders because doing so could unintentionally exclude patients with LSR and because certain disorders could be controlled for later using PSM.

Diagnostic codes corresponding to LSR and nonradicular LBP were adapted from previous studies.6,10,17,47 Definitions of codes were identified using a website and converted between ICD-10 and ICD-9 by this site's online conversion tool.42 Our LSR cohort was intended to be specific by only including codes that indicate nerve root involvement related to the low back, particularly those associated with discogenic radiculopathy, the most common cause of LSR.19

Certain ICD codes are neutral with regards to the presence of nerve root involvement, such as those for LDH or spondylolisthesis, which can cause LBP with or without LSR.39 To account for this ambiguity, this study did not include patients with these diagnoses unless they had a specific LSR diagnosis (Supplementary material, Table 2, available at

The nonradicular LBP cohort included patients with the code for “low back pain” (ICD-9: 724.4). Diagnoses specific to LSR as well as those that could relate to LSR, such as sciatica, and disk displacement (Supplementary material, Table 3, available at were excluded from the nonradicular LBP cohort, in addition to the exclusions for both cohorts (Supplementary material, Table 1, available at Spondylolisthesis and spondylolysis were also excluded from the nonradicular LBP cohort as some research has found an association between these conditions and LSR.27,39

2.5. Variables

For the purposes of this study, constipation was defined using a composite outcome of constipation diagnoses and laxative medications (Supplementary material, Table 4, available at over a 1-year follow-up window. This phenotype was developed using a previous study as a starting point45; however, our study differs by not using the code for IBS as an outcome. The IBS code was removed for our study because the IBS patterns of “mixed diarrhea and constipation” or “diarrhea only” are more common than “constipation only.31” Although ICD-10 provides more detailed codes specifying the presence or absence of constipation and diarrhea associated with IBS, general equivalence mapping is performed automatically in the TriNetX platform.

Our study included patients taking laxatives in the constipation outcome by using the drug class GA200 in the Veterans Health Administration National Drug File (VANDF).20 This strategy was intended to increase the sensitivity for detecting patients with constipation, considering some patients using laxatives may not have been diagnosed as such in the EMR, as a previous study found that many patients do not discuss constipation with their medical provider.46 Another related benefit is that this strategy may identify patients taking laxatives for constipation related to a “constipation only” form of IBS, which was not included in the constipation outcome.

2.6. Potential confounders

This study used PSM, a method of balancing covariates between cohorts recommended to reduce bias in retrospective observational studies.12 Covariates present within the year preceding the index date of LBP diagnosis could be factored into PSM. Covariates were specified a priori based on their association with constipation as described in a literature review of common causes of constipation.9 These included medications used for LSR and endocrine, neurologic, and opioid use disorders. Irritable bowel syndrome was propensity matched to improve the validity of results because IBS can cause diarrhea or constipation.31

Neurological disorders were matched by choosing diagnoses within the ICD-10 category G00-G99 with the greatest between-cohort baseline difference. This was intentional because controlling for all neurological conditions (eg, neuralgia) could unintentionally exclude patients with LSR. Conversely, the entire ICD-10 category “endocrine, nutritional, and metabolic diseases” (E00-E89) was controlled for as many of these disorders can cause constipation, such as hypothyroidism and diabetes.9

Central nervous system medications as listed in the VANDF class CN10020 were factored into PSM. These included but were not limited to analgesics, anesthetics, sedatives or hypnotics, anticonvulsants, antidepressants, antipsychotics, lithium salts, and central nervous system stimulants (Supplementary material, Table 5, available at The drug class for musculoskeletal medications (MS000) was likewise added to PSM. We did not control for corticosteroid medications, which have less evidence of an association with constipation.

2.7. Study size

A sample size of 988 was calculated using G*Power and z-tests for logistic regression, with a power of 0.95, α error of 0.05, and assuming a normal distribution. A value of 0.9 was used for R2 to estimate a high level of interaction between covariates: the probability of constipation given the null hypothesis was 0.27, the higher estimate of the prevalence of constipation in adults from a previous review,28 and the probability of constipation given the alternative hypothesis was 0.45, the incidence of constipation in patients with sciatica in a previous study.1

2.8. Statistical methods

TriNetX uses logistic regression to calculate propensity scores for patients in each cohort, and a matching ratio of 1:1 using greedy nearest neighbor matching, with a caliper of 0.01 pooled standard deviations. Baseline characteristics were compared using an independent samples t test for continuous variables and a Pearson χ2 test for categorical variables (presented as percentages and frequencies, Table 1). Missing data were examined by comparing the average facts per patient because a similar data density between cohorts suggests a minimal effect of documentation bias.35

Table 1 - Baseline characteristics.
Characteristic Before matching After matching
LSR Nonradicular LBP P LSR Nonradicular LBP P
N 503,760 1,071,498 503,062 503,062
Age 36.9 ± 8.22 33.9 ± 9.02 <0.001 36.9 ± 8.21 37 ± 8.25 0.0139
 Female 288,696 (57.31%) 622,506 (58.10%) <0.001 288,269 (57.30%) 288,104 (57.27%) 0.7394
 Male 214,918 (42.66%) 448,640 (41.87%) <0.001 214,648 (42.67%) 214,794 (42.70%) 0.768545
 Black 87,454 (17.36%) 233,811 (21.82%) <0.001 87,452 (17.38%) 85,734 (17.04%) <0.001
 White 327,023 (64.92%) 630,010 (58.80%) <0.001 326,365 (64.88%) 328,740 (65.35%) <0.001
 Asian 10,972 (2.18%) 27,382 (2.56%) <0.001 10,971 (2.18%) 10,598 (2.11%) 0.0102
 American Indian 2,350 (0.47%) 4,657 (0.44%) 0.0050 2,347 (0.47%) 2,171 (0.43%) 0.0086
 Pacific Islander 895 (0.18%) 1,916 (0.18%) 0.8731 894 (0.18%) 836 (0.17%) 0.1628
 Hispanic/Latino 41,904 (8.32%) 102,477 (9.56%) <0.001 41,895 (8.33%) 41,250 (8.20%) 0.0195
 Not Hispanic/Latino 279,116 (55.41%) 573,973 (53.57%) <0.001 278,619 (55.39%) 280,955 (55.85%) <0.001
Conditions (ICD-10)
 Cerebral infarction (I63) 1,564 (0.31%) 2,763 (0.26%) <0.001 1,562 (0.31%) 1,306 (0.26%) <0.001
 Episodic and paroxysmal disorders (G40-G47) 92,053 (18.27%) 163,633 (15.27%) <0.001 91,725 (18.23%) 91,015 (18.09%) 0.0663
 Multiple sclerosis (G35) 2,191 (0.44%) 3,345 (0.31%) <0.001 2,177 (0.43%) 1,889 (0.38%) <0.001
 Myopathy (G72.9) 323 (0.06%) 482 (0.05%) <0.001 321 (0.06%) 246 (0.05%) 0.0016
 Cerebral palsy (G80) 454 (0.09%) 1,422 (0.13%) <0.001 454 (0.09%) 284 (0.06%) <0.001
 Irritable bowel syndrome (K58) 6,718 (1.33%) 9,988 (0.93%) <0.001 6,654 (1.32%) 6045 (1.20%) < 0.001
 Endocrine, nutritional, and metabolic diseases (E00-E89) 131,502 (26.10%) 231,202 (21.58%) <0.001 131,072 (26.06%) 130,497 (25.94%) 0.1912
 Autoimmune thyroiditis (E06.3) 1,947 (0.39%) 3,265 (0.31%) <0.001 1,937 (0.39%) 1,712 (0.34%) <0.001
 Thyrotoxicosis (E05.0) 1,290 (0.26%) 2,248 (0.21%) <0.001 1,285 (0.26%) 1,081 (0.22%) <0.001
 Cushing syndrome (E24) 201 (0.04%) 334 (0.03%) 0.0055 201 (0.04%) 175 (0.04%) 0.1798
 Opioid-related disorders (F11) 6,464 (1.28%) 11,956 (1.12%) <0.001 6,449 (1.28%) 5,905 (1.17%) <0.001
Medications (VANDF)
 CNS medications (CN000) 248,421 (49.31%) 454,121 (42.38%) <0.001 247,723 (49.24%) 246,508 (49.00%) 0.0153
 Opioid analgesics (CN101) 170,884 (33.92%) 293,119 (27.36%) <0.001 170,272 (33.85%) 169,086 (33.61%) 0.0123
 Nonopioid analgesics (CN103) 157,423 (31.25%) 283,628 (26.47%) <0.001 156,958 (31.20%) 156,009 (31.01%) 0.0409
 Nonsteroidal anti-inflammatory analgesics (CN104) 106,775 (21.20%) 207,688 (19.38%) <0.001 106,629 (21.20%) 105,570 (20.99%) 0.0096
 Sedatives/hypnotics (CN300) 98,183 (19.49%) 155,912 (14.55%) <0.001 97,729 (19.43%) 97,242 (19.33%) 0.2193
 Anticonvulsants (CN400) 56,364 (11.19%) 69,959 (6.53%) <0.001 55,679 (11.07%) 51,913 (10.32%) <0.001
 Musculoskeletal medications (MS000) 201,907 (40.08%) 346,234 (32.31%) <0.001 201,211 (40.00%) 199,930 (39.74%) 0.0091
P values < 0.05 in bold.
CNS, central nervous system, LBP, low back pain, LSR, lumbosacral radiculopathy, VANDF, Veterans Health Administration National Drug File.

3. Results

3.1. Participants

A large sample size was identified for each cohort. Before PSM, there were 503,760 patients in the LSR cohort and 1,071,498 in the nonradicular LBP cohort. After PSM, there were 503,062 patients in each cohort. The LSR cohort had a significantly greater age, a greater incidence of most comorbidities except for muscular dystrophy and IBS, and greater utilization of medications (Table 1).

To safeguard PHI and improve performance times, TriNetX obfuscates patient counts within their “Explore Cohort tool,” in which results are limited to ∼10,000 patients per HCO. Analyses of baseline characteristics and outcomes include all patients meeting the study selection criteria. In accordance with the limitation of exploring the entire cohorts in greater detail, an analysis of 358,287 of patients randomly selected from the LSR cohort was performed to characterize the frequency of included diagnoses. The most common diagnoses in the LSR cohort were radiculopathy in the lumbar region (ICD-10: M54.16), which was present in 45% of patients, followed by sciatica (ICD-10: M54.3, 38%), lumbago with sciatica (ICD-10: M54.4, 37%), and radiculopathy in the lumbosacral region (ICD-10: M54.17, 27%). Lumbosacral root disorders, not elsewhere classified (ICD-10: G54.4) and radiculopathy, sacral and sacrococcygeal region (ICD-10: M54.18) were uncommon and found in <1% of the LSR cohort. The distribution of diagnoses suggested a successful identification of the target population for our LSR cohort.

3.2. Descriptive data

The number of data points between cohorts were compared, showing a high average number of facts per patient (LSR 812 and nonradicular 698), which was similar between cohorts, suggesting a minimal effect of documentation bias or missing information. Because of the large sample size and large prematching differences, there were statistically significant differences in certain variables postmatching. A visual diagnostic showed that the propensity scores between cohorts were well-balanced despite these differences (Supplementary material, Figure 2, available at

3.3. Key results

The odds of constipation were equivalent between the LSR and nonradicular LBP cohorts (LSR: 10.8%, nonradicular LBP: 10.9%; odds ratio [OR] [confidence interval] = 0.99 [0.98-1.0], P = 0.251). The odds remained nearly identical after propensity score matched analysis (LSR: 10.8%, nonradicular LBP: 11.1%; OR [confidence interval] = 0.98 [0.97-0.99], P = 0.003). Although the P value after PSM indicated statistical significance of the OR, the magnitude of the OR remained nearly equal to the null value of 1.0.

3.4. Sensitivity analysis

Although sensitivity analysis was planned for this study using the E-value,44 this was not applicable because of the lack of an identified increase in odds of constipation in the LSR cohort.

4. Discussion

This purpose of this retrospective study was to investigate the possible association between LSR and constipation using a real-world national database, with the hypothesis that adults with LSR have increased odds of developing constipation compared with those with nonradicular LBP. To the best of our knowledge, this study was the largest of its kind including 503,062 patients in each LSR and nonradicular LBP cohorts using the large TriNetX platform.

Initial analyses indicated that patients with LSR had equivalent odds of developing constipation than nonradicular LBP before PSM (OR 0.99), and this was unchanged (OR 0.98) after matching for demographics, comorbidities, and medications. This suggests that there is no clinically significant association between LSR and constipation compared with nonradicular LBP.

This study does not rule out an association between constipation and LBP in general, given that about 11% of patients in each cohort developed constipation within the year after diagnosis. Although this rate could be reflective of the high prevalence of constipation in the general population,28 our study only analyzed incident cases of constipation over one year. The percentage of patients developing constipation in both LSR and nonradicular LBP cohorts could be congruent with previous research that identified an association between LBP and constipation.33,49

The lack of an independent association between LSR and constipation suggests that the spinal defecatory reflex is not diminished to a significant degree or is not required to maintain colonic motility even if abolished in these patients. It is possible that the remaining functional lumbosacral roots permit continued function of this reflex or that the enteric nervous system maintains colonic motility despite the loss of the spinal defecatory reflex.

The results of this study should be contrasted with previous studies reporting an association between LBP and urinary incontinence.16,51 Although patients with LSR are more at risk of urinary incontinence,16 our study shows this relationship does not hold true for constipation. It is possible that there is a lower threshold to urinary incontinence in LSR, whereas bowel dysfunction is only affected in those with more severe forms of LSR such as CES. This explanation relates to neurophysiological differences because the bladder seems to be more reliant on its spinal innervation as compared to the colon.

Given that constipation may develop subsequent to either LSR or nonradicular LBP diagnoses with similar frequency, clinicians should be mindful of the constipating effects of medications used to treat pain in patients with LBP such as opioids, sedatives, and anticonvulsants. When treating LBP, clinicians may prescribe medications with a reduced risk of constipation, or over a shorter duration, or recommend evidence-based nonpharmacologic therapies.

One previous study found that increasing pain severity is positively associated with constipation severity.2 This may explain the equivalent rate of constipation in both LSR and nonradicular LBP cohorts, given both conditions are inherently painful. This previously identified association underscores the importance of treating pain in patients with low back conditions with nonpharmacologic interventions rather than opioid medications, which can increase the risk of constipation.22 Indeed, a recent clinical practice guideline from the American College of Physicians provided a strong recommendation for superficial heat, massage, acupuncture, or spinal manipulation for acute or subacute LBP.25

Strategies to promote movement and exercise could benefit patients with LBP and constipation because physical inactivity is associated with both conditions. Although the aforementioned treatments for pain may enable patients to be more active, it may be beneficial to directly encourage evidence-based exercise therapies for LBP, such as tai chi or yoga, or refer patients for physical therapy.25 In general, lifestyle factors such as regular walking or exercise are protective against LBP.30

Constipation in those with LBP is an important interdisciplinary consideration. Providers treating the spine should communicate with primary care or gastroenterology providers to facilitate an optimal outcome for their patients because constipation has unwanted sequelae such as hemorrhoids and diverticular disease. In exchange, management of constipation may reduce LBP-related exacerbations because straining, eg, during elimination, often exacerbates discogenic LSR.38

5. Limitations

Although we accounted for differences between cohorts using PSM, there are numerous residual confounding variables such as those related to pain severity,2 physical activity level,53 socioeconomic status, stress, hormonal changes, water intake,53 and dietary variables, which were not available in our data set. Another potential confounder is pelvic floor dysfunction, which has a limited association to LBP but strong association with constipation.9 However, this diagnosis is poorly represented in the ICD coding system, making it difficult to account for in our study.

We were unable to determine the severity of constipation in each cohort. Data from a patient-reported outcome index for constipation such as the Constipation Scoring System was unavailable in the TriNetX platform, and if present, could have enabled this information to be known. Although the incidence of constipation was similar between cohorts, it is possible that the severity of constipation differed. In addition, data regarding the dose and duration of laxative medications were unavailable in TriNetX. As laxatives were part of the composite outcome for constipation, this missing information could have affected study results.

Some patients could have been misclassified in the EHR. This could happen if patients received an incorrect diagnosis or older, inactive comorbidities were copied into the current chart. Prevalent users could have been included if they received care outside of the HCOs included in TriNetX and had a previous history of constipation and LBP. In addition, some patients could have been lost to follow-up. Although there is no guarantee that all included patients had incident LBP, a washout period to exclude previous LBP minimized this bias by including patients at a similar time point coinciding with the index date of LBP diagnosis.

We were unable to determine the severity of LSR. Patient-reported outcome assessments, imaging results, and pain scores are unavailable in TriNetX. Although we attempted to create a uniform cohort of patients with LSR by excluding those with CES, the remaining patients could have differed with regards to having mild, moderate, or severe nerve root compression and corresponding neurologic deficits.

Patients with LSS were excluded from this study. Although LSS commonly causes LSR, it may have significant clinical and pathophysiological differences from discogenic LSR. As the onset of neurologic deficits is more insidious, a longer follow-up may be required. In addition, comorbidities such as dementia would need to be considered within the study design. This study could be repeated with a focus on patients age 50 and older with LSS. Finally, our study design did not permit us to investigate constipation as a result of CES, in which lumbosacral nerve roots are more severely injured. Most patients with LSR have an injury to one or 2 nerve roots, usually L5 or S1,21 whereas those with CES have multiple nerve roots injured and are more likely to have involvement of the sacral roots, which are important for bowel and bladder function.41 Further research could compare constipation in patients with LSR to those with CES.

6. Conclusions

The results of this study refute the hypothesis that patients with LSR have increased odds of constipation compared with those with nonradicular LBP, given that the results showed equivalent odds of this outcome using a large propensity-matched sample. Although clinicians should be vigilant to recognize constipation in patients with LBP, they should understand that LSR is unlikely to be directly causative. Numerous variables contribute to constipation which may be present in LBP in general, including increasing pain severity, physical inactivity, and medication side effects. In addition, future researchers should consider using TriNetX so that real-world EHR data can inform clinical practice.

Appendix A. Supplemental digital content

Supplemental digital content associated with this article can be found online at


R.J. Trager reports he has received book royalties as the author of 2 texts on the topic of sciatica. S.R.S. Mok reports personal fees from CONMED and Medtronic, outside the submitted work.

This study was deemed Not Human Subjects Research by the University Hospitals Institutional Review Board.

R.J. Trager, S.R.S. Mok, and J.A. Dusek conceived of and designed the study. R.J. Trager, K.J. Schlick, and J.A. Perez performed data collection. R.J. Trager, S.R.S. Mok, K.J. Schlick, J.A. Perez, and J.A. Dusek were responsible for data analysis and interpretation. All authors contributed to, critically revised, and approved of the final manuscript. We would like to thank Sarah E. Prosak, DC, at University Hospitals, for suggestions regarding the clinical impact of the study findings.


[1]. Ali M, Shukla VD, Dave AR, Bhatt NN. A clinical study of Nirgundi Ghana Vati and Matra Basti in the management of Gridhrasi with special reference to sciatica. Ayu 2010;31:456–60.
[2]. Arai Y-C, Shiro Y, Funak Y, Kasugaii K, Omichi Y, Sakurai H, Matsubara T, Inoue M, Shimo K, Saisu H, others. The association between constipation or stool consistency and pain severity in patients with chronic pain. Anesthesiol Pain Med 2018;8:e69275.
[3]. Bourgeois J, Elseviers MM, Azermai M, Van Bortel L, Petrovic M, Vander Stichele RR. Benzodiazepine use in Belgian nursing homes: a closer look into indications and dosages. Eur J Clin Pharmacol 2012;68:833–44.
[4]. Brierley SM, Hibberd TJ, Spencer NJ. Spinal afferent innervation of the colon and rectum. Front Cell Neurosci 2018;12:467.
[5]. Chang J, Locke GR, Schleck C, Zinsmeister A, Talley N. Risk factors for chronic constipation and a possible role of analgesics. Neurogastroenterol Motil 2007;19:905–11.
[6]. Cherkin DC, Deyo RA, Volinn E, Loeser JD. Use of the international classification of diseases (ICD-9-CM) to identify hospitalizations for mechanical low back problems in administrative databases. Spine 1992;17:817–25.
[7]. Chou R, Deyo R, Friedly J, Skelly A, Weimer M, Fu R, Dana T, Kraegel P, Griffin J, Grusing S. Systemic pharmacologic therapies for low back pain: a systematic review for an American College of physicians clinical practice guideline. Ann Intern Med 2017;166:480–92.
[8]. Chou R, Qaseem A, Snow V, Casey D, Cross JT, Shekelle P, Owens DK. For the clinical efficacy assessment subcommittee of the American College of physicians and the American College of physicians/American pain society low back pain guidelines panel*. Diagnosis and treatment of low back pain: a joint clinical practice guideline from the American College of physicians and the American pain society. Ann Intern Med 2007;147:478.
[9]. Forootan M, Bagheri N, Darvishi M. Chronic constipation: a review of literature. Medicine (Baltimore) 2018;97:e10631.
[10]. Fritz JM, Kim J, Dorius J. Importance of the type of provider seen to begin health care for a new episode low back pain: associations with future utilization and costs. J Eval Clin Pract 2016;22:247–52.
[11]. Giardina JC, Cha T, Atlas SJ, Barry MJ, Freiberg AA, Leavitt L, Marques F, Sepucha K. Validation of an electronic coding algorithm to identify the primary indication of orthopedic surgeries from administrative data. BMC Med Inform Decis Mak 2020;20:1–10.
[12]. Gokhale M, Stürmer T, Buse JB. Real-world evidence: the devil is in the detail. Diabetologia 2020;63:1694–1705.
[13]. Hippocrates CoxeJR. Galen. The writings of Hippocrates and Galen. Philadelphia: Lindsay and Blakiston, 1846. Available at: Accessed December 4, 2014.
[14]. Jönsson B, Strömqvist B. Influence of age on symptoms and signs in lumbar disc herniation. Eur Spine J 1995;4:202–5.
[15]. Kamel JT, D'Souza WJ, Cook MJ. Severe and disabling constipation: an adverse effect of pregabalin. Epilepsia 2010;51:1094–6.
[16]. Kaptan H, Kulaksızoğlu H, Kasımcan Ö, Seçkin B. The association between urinary incontinence and low back pain and radiculopathy in women. Open Access Maced J Med Sci 2016;4:665.
[17]. Kazis LE, Ameli O, Rothendler J, Garrity B, Cabral H, McDonough C, Carey K, Stein M, Sanghavi D, Elton D, others. Observational retrospective study of the association of initial healthcare provider for new-onset low back pain with early and long-term opioid use. BMJ Open 2019;9:e028633.
[18]. Konstantinou K, Dunn KM. Sciatica: review of epidemiological studies and prevalence estimates. Spine 2008;33:2464–72.
[19]. Konstantinou K, Dunn KM, Ogollah R, Vogel S, Hay EM, others. Characteristics of patients with low back and leg pain seeking treatment in primary care: baseline results from the ATLAS cohort study. BMC Musculoskelet Disord 2015;16:332.
[20]. Lincoln M. Veterans health administration national drug file—summary | NCBO BioPortal. n.d. Available at: Accessed April 3, 2021.
[21]. Nafissi S, Niknam S, Hosseini SS. Electrophysiological evaluation in lumbosacral radiculopathy. Iran J Neurol 2012;11:83–6.
[22]. Pappagallo M. Incidence, prevalence, and management of opioid bowel dysfunction. Am J Surg 2001;182:S11–18.
[23]. Pliny the E, Bostock J, Riley HT, Henry T. The natural history of Pliny. London, 1856. Available at: Accessed January 16, 2015.
[24]. Podnar S. Cauda equina lesions as a complication of spinal surgery. Eur Spine J 2010;19:451–7.
[25]. Qaseem A, Wilt TJ, McLean RM, Forciea MA. Noninvasive treatments for acute, subacute, and chronic low back pain: a clinical practice guideline from the American College of physicians. Ann Intern Med 2017;166:514–30.
[26]. Rao SS, Meduri K. What is necessary to diagnose constipation? Best Pract Res Clin Gastroenterol 2011;25:127–40.
[27]. Sairyo K, Sakai T, Amari R, Yasui N. Causes of radiculopathy in young athletes with spondylolysis. Am J Sports Med 2010;38:357–62.
[28]. Schmidt FMQ, de Gouveia Santos VLC. Prevalence of constipation in the general adult population: an integrative review. J Wound Ostomy Continence Nurs 2014;41:70–6.
[29]. Shapiro S. Medical realities of cauda equina syndrome secondary to lumbar disc herniation. Spine 2000;25:348–52.
[30]. Shiri R, Coggon D, Falah-Hassani K. Exercise for the prevention of low back pain: systematic review and meta-analysis of controlled trials. Am J Epidemiol 2018;187:1093–101.
[31]. Singh P, Staller K, Barshop K, Dai E, Newman J, Yoon S, Castel S, Kuo B. Patients with irritable bowel syndrome-diarrhea have lower disease-specific quality of life than irritable bowel syndrome-constipation. World J Gastroenterol 2015;21:8103.
[32]. Sithamparanathan K, Sadera A, Leung L. Adverse effects of benzodiazepine use in elderly people: a meta-analysis. Asian J Gerontol Geriatr 2012;7:107–11.
[33]. Smith MD, Russell A, Hodges PW. How common is back pain in women with gastrointestinal problems? Clin J Pain 2008;24:199–203.
[34]. Smith-Edwards K, Edwards B, Albers K, Najjar S, Davis B. 225Extrinsic colon afferents influence enteric neuron activity and colon motility by engaging spinal reflexes. J Pain 2019;20:S31.
[35]. Stapff MP. Using real world data to assess cardiovascular outcomes of two antidiabetic treatment classes. World J Diabetes 2018;9:252.
[36]. Stienen MN, Smoll NR, Hildebrandt G, Schaller K, Tessitore E, Gautschi OP. Constipation after thoraco-lumbar fusion surgery. Clin Neurol Neurosurg 2014;126:137–42.
[37]. Stynes S, Konstantinou K, Dunn KM. Classification of patients with low back-related leg pain: a systematic review. BMC Musculoskelet Disord 2016;17:226.
[38]. Stynes S, Konstantinou K, Ogollah R, Hay EM, Dunn KM. Clinical diagnostic model for sciatica developed in primary care patients with low back-related leg pain. PLoS One 2018;13:e0191852.
[39]. Suri P, Boyko EJ, Goldberg J, Forsberg CW, Jarvik JG. Longitudinal associations between incident lumbar spine MRI findings and chronic low back pain or radicular symptoms: retrospective analysis of data from the longitudinal assessment of imaging and disability of the back (LAIDBACK). BMC Musculoskelet Disord 2014;15:1.
[40]. Tamburrelli F, Genitiempo M, Bochicchio M, Donisi L, Ratto C. Cauda equina syndrome: evaluation of the clinical outcome. Eur Rev Med Pharmacol Sci 2014;18:1098–105.
[41]. Tarulli AW, Raynor EM. Lumbosacral radiculopathy. Neurol Clin 2007;25:387–405.
[42]. The Web's free 2021 ICD-10-CM/PCS medical coding reference. n.d. Available at: Accessed January 16, 2021.
[43]. Topaloglu U, Palchuk MB. Using a federated network of real-world data to optimize clinical trials operations. J Clin Cancer Inform 2018;2:1–10.
[44]. Trager RJ, Mok SRS, Schlick KJ, Dusek JA. Association between radicular low back pain and constipation: A retrospective cohort study protocol, 2021. Available at:
[45]. Trinkley KE, Sill BE, Porter K, Nahata MC. Prescribing patterns for outpatient treatment of constipation, irritable bowel syndrome-related constipation, and opioid-induced constipation: a retrospective cross-sectional study. J Manag Care Spec Pharm 2015;21:1077–87.
[46]. Vallerand AH, Hendry S, Baldys E, Hu Y, Datto C. Analysis of patient–provider interactions regarding the burden and treatment of opioid-induced constipation in adults with chronic noncancer pain. Pain Med 2019;20:889–96.
[47]. Vogt MT, Kwoh CK, Cope DK, Osial TA, Culyba M, Starz TW. Analgesic usage for low back pain: impact on health care costs and service use. Spine 2005;30:1075–81.
[48]. Von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Ann Intern Med 2007;147:573–7.
[49]. Walden AL, Salsbury SA, Reed WR, Lawrence DJ. Bladder and bowel symptoms among adults presenting with low back pain to an academic chiropractic clinic: results of a preliminary study. J Chiropr Med 2014;13:178–87.
[50]. Watanabe K, Sekiguchi M, Yonemoto K, Nikaido T, Kato K, Otani K, Yabuki S, Kakuma T, Kikuchi S, Konno S, others. Bowel/bladder dysfunction and numbness in the sole of the both feet in lumbar spinal stenosis–a multicenter cross-sectional study. J Orthop Sci 2017;22:647–51.
[51]. Welk B, Baverstock R. Is there a link between back pain and urinary symptoms? Neurourol Urodyn 2020;39:523–32.
[52]. Younes M, Béjia I, Aguir Z, Letaief M, Hassen-Zrour S, Touzi M, Bergaoui N. Prevalence and risk factors of disk-related sciatica in an urban population in Tunisia. Joint Bone Spine 2006;73:538–42.
[53]. Yurtdaş G, Acar-Tek N, Akbulut G, Cemali Ö, Arslan N, Beyaz Coşkun A, Zengin FH. Risk factors for constipation in adults: a cross-sectional study. J Am Coll Nutr 2020;39:713–19.

Low back pain; Constipation; Radiculopathy; Electronic medical records; Confounding variables; Propensity score

Supplemental Digital Content

Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of The International Association for the Study of Pain.