Steven P. Cohen, Conner Nguyen, Shruti G. Kapoor, Victoria C. Anderson-Barnes, Leslie Foster, Cynthia Shields, Brian McLean, Todd Wichman, Anthony Plunkett. Back Pain During WarAn Analysis of Factors Affecting Outcome. Arch Intern Med. 2009;169(20):1916–1923. doi:10.1001/archinternmed.2009.380
Back pain is the leading cause of disability in the world, but it is even more common in soldiers deployed for combat operations. Aside from battle injuries and psychiatric conditions, spine pain and other musculoskeletal conditions are associated with the lowest return-to-unit rate among service members medically evacuated out of Operations Iraqi and Enduring Freedom.
Demographic, military-specific, and outcome data were prospectively collected over a 2-week period at the Deployed Warrior Medical Management Center in Germany on 1410 consecutive soldiers medically evacuated out of theaters of combat operations for a primary diagnosis pertaining to back pain between 2004 and 2007. The 2-week period represents the maximal allowable time an evacuated soldier can spend in treatment before disposition (ie, return to theater or evacuate to United States) is rendered. Electronic medical records were then reviewed to examine the effect a host of demographic and clinical variables had on the categorical outcome measure, return to unit.
The overall return-to-unit rate was 13%. Factors associated with a positive outcome included female sex, deployment to Afghanistan, being an officer, and a history of back pain. Trends toward not returning to duty were found for navy and marine service members, coexisting psychiatric morbidity, and not being seen in a pain clinic.
The likelihood of a service member medically evacuated out of theater with back pain returning to duty is low irrespective of any intervention(s) or characteristic(s). More research is needed to determine whether concomitant treatment of coexisting psychological factors and early treatment “in theater” can reduce attrition rates.
There is a widespread misconception that battle-related injuries are the major source of unit declension in the military. Yet, since statistics have been kept, the main causes of soldier attrition in modern war have never been injuries sustained in combat. In World War I, respiratory illness and infectious disease were the leading causes of hospital admission, with non–battle-related injuries (NBIs) ranking fourth. In World War II and the Korean conflict, NBIs ranked third as a cause of unit dissolution. By the time of the Vietnam War, NBIs had become the leading basis for soldier attrition, where they have remained ever since.1
These facts should come as no surprise to those familiar with the logistics of modern military operations. Since the end of the Vietnam conflict, the strategic objectives of most modern military conflicts have shifted from rampant destruction to the preservation of infrastructure and culture, minimizing collateral damage, and constructive engagement with local nationals. This change in emphasis has resulted in a parallel increase in the ratio of support to combat troops, which now approaches 7:1.
Similar to a civilian cohort,2 low back pain (LBP) is one of the principal reasons soldiers seek medical attention in mature theaters of operation.3 The factors associated with back pain outcome in a forward-deployed setting, which for a deployed soldier can be condensed to “return to duty,” may be akin to the civilian sector, whereby coexisting psychopathologic and psychosocial stressors, low job satisfaction, and other work-related factors, such as lack of autonomy and an inadequate support structure, are major determinants of disability.4- 7
The profound role nonanatomical factors play in unit preservation are perhaps best exemplified by comparisons between return-to-duty rates for LBP and other NBIs in forward-deployed and rear-echelon military treatment facilities (MTFs). In an observational study by Cohen et al8 conducted in level IV MTF pain clinics in and outside the continental United States, less than 2% of soldiers medically evacuated out of theater returned to their units. However, when soldiers with comparable diagnoses were treated in a pain clinic set up in a level III MTF (ie, combat support hospital), 95% remained with their units.3 One factor that may have contributed to the large disparity in outcomes was the relative absence of confounding psychosocial variables in the soldiers treated in combat support hospitals.4 For example, soldiers treated in-theater may have been more vested in their mission and less likely to have a coexisting psychopathologic condition. The observation that return-to-unit rates for pain stratified by treatment level bear a striking resemblance to combat stress, for which the return-to-unit rates exponentially decline the further away from their unit the soldier is treated, rather than an acute medical condition such as renal stones, for which the site of treatment has little effect on outcome, lends credence to this theory.
In November 2002, the US Medical Command set up the Deployed Warrior Medical Management Center (DWMMC) to coordinate medical evacuations from Operations Iraqi Freedom (OIF) and Enduring Freedom (OEF) to Landstuhl, Germany, and optimize the case management of wounded soldiers. In the ensuing years, DWMMC has taken on a number of additional roles, one of which is to serve as a database for all medically evacuated personnel. On the basis of International Classification of Diseases, Ninth Revision (ICD-9) diagnostic codes, spine pain was identified as the fifth leading cause of medical evacuation from OIF and OEF to the level III MTF in Landstuhl, Germany. Factors associated with back pain outcome have been well delineated in a civilian cohort4- 7 but to our knowledge have never been determined in a military setting. The purpose of this study was to identify which variables are associated with return to unit in soldiers with a primary diagnosis of back pain medically evacuated out of a combat theater.
Permission to conduct this study was granted by the US Medical Command and the Department of Clinical Investigation at Walter Reed Army Medical Center, Washington, DC. The lists of soldiers medically evacuated out of forward-deployed units for OIF, OEF, and “other” operations between 2004 and 2007 were obtained from a prospective database actively maintained in Landstuhl, Germany, by the DWMMC. Diagnoses were conferred by the physician arranging the medical evacuation and recorded based on the ICD-9. Among soldiers with multiple diagnoses, only the primary diagnosis (ie, the cause of medical evacuation) is coded (Table 1). Consequently, each patient listing contains only a single diagnosis. Other data collected in the database included age, sex, rank, date of evacuation, whether the injury was incurred during battle, deployment mission (ie, OIF, OEF, or “other”), and disposition (ie, return to unit or transfer to an MTF in the continental United States).
Electronic medical records were subsequently reviewed to obtain data on variables suspected of influencing return to duty. These included whether the predominant symptoms were axial or radicular, region (lumbar or thoracic), corresponding pathologic magnetic resonance imaging (MRI) findings, mechanism of injury, presence of a coexisting psychiatric condition, smoking history, treatment modality (eg, interventional or conservative), and whether the soldiers were seen by a pain specialist in theater or Germany (ie, prior to final outcome). Because disposition was annotated alongside demographic data used to access electronic records, data mining was not blinded to outcome.
The operation wherein the injury occurred (eg, OIF or OEF) was tabulated directly from the database, which was sourced by the soldier's deployment orders. In addition to the principal mission in Afghanistan, OEF included subordinate operations in Kyrgyzstan, Horn of Africa, Trans-Sahara, Philippines, Europe, and Pankisi Gorge. The OIF orders included not only deployments to Iraq proper, but support missions involving the Arabian Peninsula (eg, Saudi Arabia, Kuwait, Qatar), Djibouti, Turkey, and the Republic of Georgia.
Battle or NBI designation was recorded directly from the DWMMC database and designated based on whether the injury was incurred during a combat mission. Rank was categorized into senior and junior levels based on the levels of responsibility afforded soldiers in each pay grade. For enlisted personnel, senior level was designated as a noncommissioned officer in the pay grade of E5 and above (eg, sergeant in the US Army or US Marines, and petty officer–second class in the US Navy). For officers, senior level was designated as a field grade or general officer (eg, major or above in the US Army or US Marines and lieutenant commander and above in the US Navy).
With regard to the electronic record review, back pain was designated as “radicular” when it extended into a leg(s), groin, or trunk in a dermatomal distribution, consistent with MRI findings. Pain was designated as “axial” when it was located predominantly in the back, with or without radiation, in a nondermatomal distribution. This information was obtained by patient history, physical examination findings (eg, straight leg raising test) when documented, and radiological imaging reports. For radicular pain, pathologic MRI findings were designated as concordant when a spinal pathologic condition was noted that could reasonably account for the patient's symptoms. For axial LBP, the corresponding MRI findings were classified as concordant when “moderate to severe” degenerative changes were present in spinal levels consistent with complaints.
The category of coexisting psychopathologic condition was recorded as positive when a patient was found to have a specific psychiatric ICD-9 diagnosis on electronic medical record review that either preceded or was related to their deployment. Treatment was designated as interventional when a spinal injection was performed either in Germany or “in-theater,” before outcome was recorded. Surgery was not considered an intervention because no soldier was surgically treated as a means to return the soldier to their unit, and no soldier who underwent spine surgery returned to theater within the allotted time frame. Finally, a positive outcome was predefined as “return to unit in full or limited capacity,” as prospectively recorded in the DWMMC database. In rare cases (<1%) when a soldier was found to have redeployed to their unit from the continental United States based on record review, this finding superseded what was noted in the database.
Statistical analyses were performed using STATA 9.1 software (StataCorp, College Station, Texas). Potential confounders to be studied were identified a priori as those variables with a P value of <.25 in χ2 analysis. Regression analysis was used to quantify the significance of various potential predictors of outcome. Because the outcome variable was binary (either positive or negative), a logistic statistical model was chosen. P < .05 was considered statistically significant, and χ2 testing was performed to select the variables related to the outcome from among the preidentified potential confounders. A model fit between these selected variables and one with all variables included was compared, with the most parsimonious model presented. Pseudo R2 was used as a marker of “goodness of fit” of the model.
The DWMMC database contained a total of 1433 patients with a primary diagnosis related to back pain. Among these, 9 were excluded for lack of outcome data (eg, failed to arrive), 13 were found to have another diagnosis (ie, neck pain or nonradicular leg pain), and 1 was a dog, leaving 1410 patients for analysis. Among the 26 700 data points, 2673 (10%) were missing. Of the missing variables, 97% stemmed from evacuations in 2004, when the military electronic medical record system had not yet been implemented in most medical facilities in Iraq and Afghanistan.
The demographic characteristics of these patients are presented in Table 2. The mean (SD) age was 33 (8.76) years. Approximately 92% of the patients were male, and 87% of the medically evacuated subjects were in the US Army, with 6% in the US Marines, 3% in the US Air Force, and 2% in the US Navy. Ninety-one percent of service members were enlisted. Forty-two percent were smokers, 28% had a coexisting psychiatric disease, and 54% had a history of back pain. Among the entire cohort, 13% (187 service members) returned to their unit.
The mechanism of injury was unknown in a majority of cases (66%). In the remaining subjects, injury was most commonly attributed to lifting objects (18%). Of the injuries, 11% were attributed to falls, 8% to driving, and 5% to road marches or fitness training. Only 5% of injuries were sustained during combat operations.
Sixty-nine percent of service members had symptoms consistent with radiculopathy. More than 67% had MRI findings that correlated with their symptoms. Owing to limited resources, only 41% of patients were referred to an interventional pain specialist. A majority of the injuries (70%) were treated conservatively, while interventional measures were used in 30% of subjects. Among the 67 patients who ended up having surgery, 5 returned to theater at a later date.
The factors associated with return to duty are detailed in Table 3 and Table 4. Although only 8% of all injuries reported were sustained by women, women were 57% more likely to return to duty after injury (95% confidence interval [CI], 25%-76%). The return-to-duty rate was 12% for men compared with 22% for women. This association was maintained in both univariate (odds ratio [OR], 0.50; 95% CI, 0.31-0.80 [P = .003]) and in the multivariate analysis after controlling for all other covariates (OR, 0.43; 95% CI, 0.25-0.75 [P = .003]). Similarly, officers were 51% (95% CI, 16%-71%) more likely to return to duty compared with enlisted personnel (P = .01). This again was true in both univariate (OR, 0.47; 95% CI, 0.30-0.74 [P = .001]) and multivariate (OR, 0.49; 95% CI, 0.29-0.84 [P = .01]) analysis. Twelve percent of enlisted personnel returned to duty compared with 23% of officers.
The operation in which injury was sustained was highly correlated with return-to-duty rates. Whereas OEF (Afghanistan) accounted for only 14% of all subjects, these patients were 77% more likely to return to duty compared with service members injured during OIF (univariate OR, 1.70; 95% CI, 1.15-2.52 [P = .008]; multivariate OR, 1.77; 95% CI, 1.12-2.81 [P = .02]). Of the subjects injured in Afghanistan, 20% returned to theater compared with only 12% of subjects injured in Iraq. A slight majority (54%) of subjects had a history of back pain. Paradoxically, these subjects were almost 50% more likely to return to their units than their counterparts who reported no previous back pain episodes (return-to-duty rates of 17% and 13%, respectively; univariate OR, 1.35; 95% CI, 0.96-1.90 [P = .08]; multivariate OR, 1.48; 95% CI, 1.01-2.18 [P = .04]).
A nonsignificant trend toward lower return-to-duty rates was observed among sailors (8% [P = .64]) and marines (10% [P = .55]). Service members in the US Air Force had an almost 70% higher return-to-duty rate (univariate OR, 1.41; 95% CI, 0.62-3.26 [P = .41]; multivariate OR, 1.72; 95% CI, 0.69-4.27 [P = .24]) compared with those serving in the US Army (18% vs 13%) and returned to their units more than twice as often as US Navy personnel (18% vs 8%). The highest return-to-unit rate was noted in nonservice members (29%). These trends held up even after controlling for potential confounding factors.
Of the evacuees, 72% did not have a coexisting psychiatric diagnosis. Although it fell shy of statistical significance, these individuals were almost 30% (univariate OR, 0.61; 95% CI, 0.40-0.91 [P = .02]; multivariate OR 0.69; 95% CI, 0.44-1.10 [P = .11]) more likely to return to work than subjects with a concomitant psychopathologic condition. Among the various subcategories of psychiatric illness, no single diagnostic group (eg, depression, combat stress) was more likely than any other to affect outcome (data not shown).
Only 41% of subjects were referred to a pain treatment center. These patients were 30% (univariate OR, 1.35; 95% CI, 0.96-1.90 [P = .08]; multivariate OR, 1.28; 95% CI, 0.88-1.86 [P = .19]) more likely to return to duty. However, no difference in return-to-duty rates was observed between those treated conservatively (15%) and those in whom interventional treatments (ie, injections) were used (15%) (univariate OR, 0.99; 95% CI, 0.68-1.43 [P = .95]; not included in multivariate model). Cause was only weakly correlated with outcome. Return-to-unit rates ranged from a high of 20% for those who sustained their injury during training to a low of 10% in service members who attributed their injury to driving. Finally, a person's age and smoking status did not appear to be related to return-to-duty rates. The mean (SD) age for soldiers who did not return to work was 32.9 (8.6) years compared with 34 (9.6) years in those who did return to work (P = .68). Similarly, both smokers and nonsmokers were equally likely to return to duty after a back pain episode (P = .94).
Perhaps the main finding in this epidemiological study is the extremely low return-to-unit rate for service members medically evacuated out of theaters of operation for LBP who were treated at the level IV MTF in Landstuhl, Germany, whose principal role in this context is something akin to a “medical way station.” Aside from true combat injuries sustained during battle, the return-to-duty rate for spinal pain and other musculoskeletal disorders is lower than for any other disease or injury category except for psychiatric illness.9
The parallels between psychiatric disorders and spine pain are striking. For psychiatric illness, the likelihood of a soldier with combat stress returning to duty declines exponentially with distance from unit.10 When combat stress symptoms are treated in forward-deployed mental health clinics co-located on major bases with combat support hospitals, 95% of service members return to their units. When treated in a combat support hospital itself or a nearby transitional facility such as Qatar or Kuwait, 75% and 50% of soldiers will return to duty, respectively. From the level IV treatment center in Germany, just under 10% of service members evacuated for a psychiatric illness return to theater. And should a combat stress patient make it all the way back to a MTF within the United States, the likelihood of that service member returning to their unit approaches 0%.
Similar proportions were found by White and Cohen3 and Cohen et al8 in a series of epidemiological studies conducted in OIF service members treated in pain clinics along the geographical spectrum of evacuation routes. Among 80 coalition forces with spine pain treated at a level III combat support hospital in Baghdad, Iraq, all patients returned to their units.3 But when a similar cohort was treated at level IV MTFs in Germany or Washington, DC, less than 2% reunited with their units. These similarities should not be surprising, considering that the major determinants for return to work and overall symptom palliation in patients with spinal pain are psychosocial.4- 6,11,12 They also suggest that the farther away an evacuee is treated from their home unit, the less likely they are to return to that unit. What is surprising is that an even stronger association was not found between a coexisting psychiatric diagnosis and return to duty.
Direct commission as an officer, female sex, deployment in support of OEF, and history of back pain were all statistically significantly associated with a positive outcome in univariate analysis. The effect of a presidential commission (eg, officer) on outcome is likely multifactorial in nature. As opposed to enlisted personnel, officers are more likely to occupy mission-essential positions and in many cases volunteer for leadership positions. Public opinion polls have shown that officers are more likely to politically support the current conflicts than enlisted personnel,13 which may be tied to job satisfaction rates and hence outcomes.14,15 For service members deployed in administrative roles such as public relations and health care, the ability to return to work in a modified, less physically stressful capacity may have contributed to superior outcomes.16- 21 Because women are supposed to occupy only noncombat roles in the US military, the latter factor may have also played a prominent role in their higher return-to-unit rates. In addition, female and male officers in command positions could have felt more internal pressure to complete their duty assignment in a male-dominated culture that emphasizes resilience.
One explanation for the impact conflict had on outcome is that at least in the early years, being deployed to Afghanistan, though more physically demanding, was widely considered to be a safer and more desirable tour of duty. But as the Iraq theater has “matured” into a nation-building mission and Afghanistan has devolved into a more dangerous one, these tenets are being challenged. In support of this hypothesis, in the past 2 years the relative number of Afghanistan evacuees has increased and the gap between the return-to-duty rates for OIF and OEF has narrowed.9
Nonuniformed service affiliation and age older than 35 years were also statistically significantly associated with a higher return-to-duty rate but only in univariate analysis. The effect nonservice affiliation had on outcome can probably be attributed to financial considerations.22,23 Although military personnel are paid a small premium for deployment to war zones, the rest of their salary is paid regardless of combat status. For private contractors, who in many cases earn salaries several times higher than service members with comparable occupational specialties, pay is usually contingent on continued deployment. With respect to the observed service differentials, one explanation is that US Air Force personnel might be granted more flexibility to return to work in a limited capacity than lower enlisted soldiers or US Marines, whose job requirements entail carrying heavy rucksacks and body armor, often in excess of 27 kg, on long road marches.
Trends toward a significant relationship were noted between outcome and previous back pain episodes and referral to a pain clinic. The influence that previous back pain episodes have on prognosis is ambiguous and likely depends on context. For example, some24 but not all25 studies have found an association between a history of back pain and negative outcome.
As pain specialists, we found it somewhat disappointing that a stronger correlation was not found between outcome and treatment at the multidisciplinary pain clinic in Landstuhl, Germany. In general, the evidence for nonsurgical procedural interventions to facilitate return to work is very weak and limited.26- 28 Although a paucity of randomized studies have critically evaluated multidisciplinary treatment programs on return-to-work rates, there is a strong consensus that this approach is beneficial.29- 31
There are several shortcomings to this study that must be addressed to appreciate the implications of these results. First, our findings were obtained in a select group of individuals faced with unique circumstances. Hence, generalizing them to civilians and/or nondeployed soldiers should be done with caution. Second, though outcomes and diagnoses were recorded prospectively, no specific set of inclusion and exclusion criteria were used, and many variables known to influence outcomes were absent from medical records. Therefore, we could not control for comorbidities (eg, inadequate coping mechanisms, duration and intensity of pain, functional limitations, fear-avoidance behavior) or rule out other confounding factors (eg, divorce, poor relationships with colleagues) that might have affected outcome.32- 34 Third, because only a single diagnosis can be annotated in the database, the numbers presented herein may underestimate the true burden of back pain during wartime. Fourth, the method for assessing psychopathologic conditions was categorical (ie, yes or no) and did not account for severity and causation, which could have weakened the association found between psychopathologic conditions and outcomes. Fifth, the method used to select these service members (eg, those already evacuated to Germany) likely skewed the sample to include those at highest risk for a negative outcome. As alluded to earlier, previous analyses have demonstrated that the further an evacuee is treated from their home station, the lower the likelihood they will return to their unit.3,10 Possible reasons for this include expectations of both health care providers and service members (ie, the “default” disposition at a combat support hospital is to return a soldier to their unit, whereas in Germany it is further evacuation), a psychological “severing” of the strong bonds with their comrades, and tacit and explicit wishes expressed by service members and their commanders (ie, a soldier who is motivated to remain “in theater” and one whose commander deems it to be mission essential to remain “in theater” is less likely to make it to Germany). Hence, these findings should not be extrapolated to individuals with back pain treated “in theater.” Finally, since a majority of the clinical variables were examined post hoc, other potential flaws include those inherent in any retrospective analysis, including a limited ability to accurately assess potential predictors of outcome measures.
In conclusion, back pain represents a major problem for soldiers deployed in combat operations and accounts for a significant proportion of unit attrition. When treated at a “remote” location, the return-to-unit rate is very low. An appreciation of those factors that affect outcome may better enable health care personnel to allocate scarce resources to those soldiers most likely to benefit from interventions.
Correspondence: Steven P. Cohen, MD, Department of Anesthesiology, Johns Hopkins School of Medicine, 550 N Broadway, Ste 301, Baltimore, MD 21029 (firstname.lastname@example.org).
Accepted for Publication: August 11, 2009.
Author Contributions:Study concept and design: Cohen. Acquisition of data: Nguyen, Kapoor, Anderson-Barnes, Foster, McLean, Wichman, and Plunkett. Analysis and interpretation of data: Cohen, Kapoor, Shields, and Plunkett. Drafting of the manuscript: Cohen, Kapoor, and Anderson-Barnes. Critical revision of the manuscript for important intellectual content: Cohen, Nguyen, Foster, Shields, McLean, Wichman, and Plunkett. Statistical analysis: Kapoor. Administrative, technical, and material support: Cohen, Nguyen, Kapoor, Anderson-Barnes, Foster, Shields, McLean, Wichman, and Plunkett
Financial Disclosure: None reported.
Funding/Support: This study was funded in part by a Congressional grant from the John P. Murtha Neuroscience and Pain Institute, Johnstown, Pennsylvania; the US Army; and the Army Regional Anesthesia & Pain Medicine Initiative, Washington, DC.
Disclaimer: The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the Department of the Army or the Department of Defense.