- Email: [email protected]
Prior depression and incident back pain among military registered nurses: A retrospective cohort study
International Journal of Nursing Studies 74 (2017) 149–154
Contents lists available at ScienceDirect
International Journal of Nursing Studies journal homepage: www.elsevier.com/locate/ijns
Prior depression and incident back pain among military registered nurses: A retrospective cohort study
MARK
⁎
D. Alan Nelsona, , Nancy Menzelb, Patricia Horohoc a b c
Department of Medicine, Stanford University School of Medicine, 450 Serra Mall, Bldg 20, Stanford, CA, 94305-2160, USA University of Nevada, Las Vegas, School of Nursing, 4505 South Maryland Parkway, Box 453018, Las Vegas, NV, 89154-3018, USA 800 North Glebe Road, Suite 300, Arlington, VA, 22203-1807, USA
A R T I C L E I N F O
A B S T R A C T
Keywords: Back ache Lower back pain Depression Epidemiology Registered nurses Occupational health Patient handling
Background: Occupational back pain rates are substantial among registered nurses, and nurses also report high rates of depression. The role of depression as a potential predictor of back pain among nurses appears understudied. Objectives: The objective of the study was to determine whether a history of depression predicted incident back pain in a population of military registered nurses when controlling for relevant risk factors. Design: We employed a retrospective cohort approach using longitudinal data in which gender-specific subject groups were followed from the beginning of duty as a registered nurse to the occurrence of an outcome, or to censoring due to completion of service or the end of available data. Participants: This study included all United States Army registered nurses who began work during 2011–2014 without evidence of prior back pain in clinical records. Methods: Data from automatically-collected medical and administrative sources were combined and used to provide 2134 person-years of observation on 1248 individuals. These data were organized at the person-month level in a panel data structure to support discrete-time multivariable logistic regression models. The models examined the relationships between prior depression, Body Mass Index, the presence of prior combat duty and selected control variables and the outcome, the incident occurrence of back pain. Results: The incidence rate of back pain was 18.6 per 100 person-years and the period prevalence was 31.7%. Prior depression was a statistically-significant predictor of incident back pain among female subjects (odds ratio [OR]: 1.75, 95% confidence interval [CI]: 1.08–2.83, P-value < 0.05). Body Mass Index of 30 kg/m2 or greater, prior combat deployments, and age 36 years or older was each associated with back pain for male and female nurses. Conclusions: The study’s findings provide the first evidence of a temporal link between antecedent depression and later back pain among female military nurses. High Body Mass Index was found to be a further, modifiable risk factor for back pain in this population.
What this paper adds
What is already known about the topic?
• Depression and back pain rates are relatively high among registered nurses and military service members. • Back pain and depression have long been associated in the litera-
ture, but there has been little work on the possible temporal relationships between these problems when controlling for other, potential contributory factors.
⁎
Corresponding author. E-mail address: [email protected] (D.A. Nelson).
http://dx.doi.org/10.1016/j.ijnurstu.2017.06.015 Received 16 December 2016; Received in revised form 22 June 2017; Accepted 23 June 2017 0020-7489/ © 2017 Elsevier Ltd. All rights reserved.
• Prior depression was associated with similar increases in the ad•
justed odds of back pain for both genders, but this association was statistically-significant among female nurses only, probably due to reduced power for males. Body Mass Index of 30 kg/m2 was associated with a similar, greater than two-fold increase in the adjusted odds of back pain among the 1248 male and female nurses studied.
International Journal of Nursing Studies 74 (2017) 149–154
D.A. Nelson et al.
1. Introduction
Table 1 Data sources leveraged to produce the panel dataset for analysis.
Registered nurses (RNs) report high rates of occupational back pain. A review of 132 international studies of workplace problems among civilian nurses (Davis and Kotowski, 2015) found that the mean, pastyear back pain among nurses working in hospitals (66 studies) was 55%. While the RN workforce faces unique health risk factors and problems, the predictors of back pain among RNs have not been fully explored. Requirements such as patient handling are known to contribute to back pain in the nursing profession (Yassi and Lockhart, 2013). However, the literature reveals little work on the relationships between mental health statuses and back pain among RNs. This deficit is noteworthy because depression rates are about twice as high among RNs working in hospitals as in the general population (Letvak et al., 2012). Depression is a common condition that is associated with back pain across multiple studies in the research literature (Hoy et al., 2010; Nicholas et al., 2011; Pinheiro et al., 2015). Depression has also been associated with a poorer prognosis in the course of back pain (Pinheiro et al., 2016). Altogether, these ideas suggest that new research addressing the relationship between depression and back pain is sorely needed in the RN population. The military provides one structured environment in which to explore the depression-back pain association. In the total United States (U.S.) Army population, back pain is a leading problem associated with lost work time (Knox et al., 2011), medical evacuation during operational deployment (Cohen et al., 2009, 2010) and disability (Lincoln et al., 2002; Patzkowski et al., 2012). Depression is also prevalent in the U.S. military (Gaderman et al., 2012; Hoge et al., 2004). However, as is also applicable to civilian RNs, we find scant past research that meaningfully addresses back pain or depression among military RNs. To address the gaps in the literature that we identified, we designed this study to examine incident back pain diagnoses among U.S. Army commissioned officers who began duty as RNs during 2011–2014. The specific aim was to assess whether a history of prior depression predicted later back pain when studying longitudinal data and controlling for other, potential back pain predictors.
Data source
Description
Defense Manpower Data Center (DMDC) Active Duty Master file Transactions file
Demographics and military service data Discharges from service
Military Health System Data Repository (MDR) Combined Ambulatory Outpatient care in military facilities Professional Encounter Record Clinical Data Repository Vital Height and weight readings from Signs outpatient encounters Standard Inpatient Data Record Inpatient care in military facilities Tricare Encounter Data, NonOutpatient care in civilian facilities Institutional Tricare Encounter Data, Inpatient care in civilian facilities Institutional Defense Training Management System (DTMS) Height/weight file Height and weight readings from biannual body composition checks Medical Operations Data System (MODS) (US Army, 2015a) Periodic Health Assessment Height and weight readings and selfreported medical problems eProfile Duty restrictions assigned by clinicians
were taken during data cleaning at the US Army Office of the Surgeon General to mitigate or eliminate missing data through cross-referencing of data elements across multiple sources before the final, de-identified dataset was provided to the University of Maryland. The sole, identifiable missing data were for Body Mass Index, addressed below. 2.2.1. Dependent variable As observed by Pinheiro et al. (2015) in a systematic review and meta-analysis of studies addressing back pain and depression, back pain definitions that result from health care-seeking or reduced functional capacity are the norm in many studies. We employed similar conceptual bases for the definition of back pain in the study population. The dichotomous dependent variable was assigned the “1” value in the person-month in which we first observed documentation of low back pain or other back ache in clinical encounter data. The International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) system remained the diagnostic system for the military as of the time of the observed data. Eligible clinical encounters or hospital admissions were those in which selected ICD-9-CM codes documented lumbar, thoracic, or regionally-unspecified back pain (724.1, 724.2, 724.5, 847.1, 847.2 or 847.9). A “1” value for the outcome variable was also assigned when back pain was entered as a clinical condition in an assigned duty restriction reflecting short- or long-term disability. Finally, back pain was defined as being present if the patient reported back pain by selecting a formatted field for this condition in a Periodic Health Assessment screening document.
2. Methods 2.1. Population and dataset The study population was established using a comprehensive, deidentified medical and administrative database on the total U.S. Army at the Center for Health Information and Decision Systems of the University of Maryland, College Park. Service members from other U.S. military branches such as the U.S. Navy were not available for study. In order to assess incident back pain after subjects began work as military RNs, our potentially-eligible subjects for the study were 1320 individuals who newly entered service as such during 2011–2014. We required that there was no evidence of prior back pain in available data from prior care and duty restrictions, rendering 72 potential subjects ineligible to produce the study population of 1248 individuals. The analytic dataset was a longitudinal panel that included military service, administrative and demographic data on the eligible subjects. The information was organized at the person-month level, with timevarying factors arranged in temporal sequence for each subject. The data sources used to produce the dataset are listed in Table 1.
2.2.2. Independent variables Increasing age may be associated with back pain (Hoy et al., 2010; Knox et al., 2011) and radiographically-verifiable spinal degeneration (de Schepper et al., 2010). Each subject's running age in years was classified using a categorical variable based on frequency distributions. Age was categorized rather than expressed as a continuous value due to non-linear effects, to be reported. The age categories were < 26 years (reference group), 26–28, 28–36, and > 36. The models also controlled for the subject’s self-reported race due to prior evidence of differences in disability rates for race-based groups (Niebuhr et al., 2011; Sikorski et al., 2012). We included a categorical variable for marital status in light of the previous association between being married and back pain in military populations (Knox et al., 2011). Subjects were classified as married, never married, or formerly married (divorced, separated or widowed).
2.2. Variables Data were obtained from official sources upon which the Department of Defense depends for critical operations. Given its ability to enforce data collection policies for service members, the demographic and most of the clinically-related variables employed in this study were remarkably free of known missing values. Additional efforts 150
International Journal of Nursing Studies 74 (2017) 149–154
D.A. Nelson et al.
outcome, departure from active military service, or the end of the available data in December 2014. When totaling the available personmonths of service time, the final dataset provided 2134 observed person-years on the 1248 subjects (mean: 20.5 months; SD: 13.3 months; range: 1–48 months.) Prior to regression analyses, we computed chi square tests for distribution differences among covariate categories for subjects with and without outcomes in order to detect potentially-useful associations. Stratified, discrete-time multivariable logistic regression models for males and females were then computed due to the frequent, widespread effect modification by gender seen in prior research on a range of health and other outcomes in the military (Nelson et al., 2016; Nelson and Kurina, 2013; Rosellini et al., 2016). Numerous, preliminary models were explored using the range of potential variable configurations that applied to each factor of interest. Final variable configurations were attained that provided maximum utility and realization of associations. This research study was approved by the Institutional Review Board at the University of Maryland and underwent secondary review by the Defense Health Agency’s Human Research Protection Office. The research has conformed to the principles embodied in the Declaration of Helsinki. All statistical analyses were conducted using Stata 14 software (StataCorp, College Station, Texas).
Body Mass Index (BMI) is an important factor due to the prior association of excess weight and back pain (Shiri et al., 2010a). We incorporated BMI as a categorical covariate due to non-linear effects to be reported. Our BMI data originated from height and weight readings taken at elective outpatient encounters, annual health assessments, and biannual administrative readings (US Army, 2013). We included a category for missing BMI data due to the presence of subjects with few or no health encounters at which readings might be taken. This was an important approach because low care frequency may correlate with less back pain. Aside from the category for missing data, the variable was organized based on the standard BMI categories (CDC, 2015). “Underweight” and “normal” ranges were combined in the reference category due to rarity of low BMI in this population. We also provided control for tobacco use due to its association with back pain (Shiri et al., 2010b). This dichotomous covariate indicated any prior self-report of tobacco use at outpatient health encounters or annual assessments. To examine the temporal relationship between prior depression and back pain in our study population, we included a rolling, dichotomous covariate indicating whether any observed evidence of depression preceded each observation. Depression events were out- or inpatient clinical diagnoses using selected ICD-9-CM codes (311, 296.2, 296.3, 309.0 or 309.1). Applicable events were also defined using documentation of back pain in Periodic Health Assessments or duty restrictions. A further binary covariate was assigned a “1” value in the longitudinal data during the last two trimesters of pregnancy, where applicable. This status was identified using required documentation of pregnancy for female soldiers in duty restrictions found in ancillary health files. We included this variable in order to control for pregnancyassociated back pain, which is a common problem for pregnant women (Vermani et al., 2010). This covariate also provided control for the military's regulation-mandated reductions in job duties during advanced pregnancy (US Army, 2011), which could conceivably reduce the risk of back pain through reduced workloads. The variable was further intended to reveal any confounding of the concurrent BMI variable, given the expected increases in BMI during later pregnancy. To control for potential variations in job duties, each nurse’s specialty area was encoded as a categorical variable. These classifications were based on the Defense Manpower Data Center records of each nurse's official primary Area of Concentration, the military term for commissioned officer job type, and additional skill identifiers. Because some Army nurses served in the military in other capacities prior to being commissioned as officers for nursing duty, such as in enlisted service or as another officer type, we identified any prior military service with a binary covariate. We also incorporated a categorical variable for the number (if any) of combat deployment experiences preceding each person-month in order to provide control for the potential effects associated with this form of military duty. This variable was included because it appeared unclear whether the mental and physical stress associated with deployments could increase the risk of back pain. An opposing theory could hold that a “healthy warrior” selection effect seen in other research (Niebuhr et al., 2011), associated with having previously been found sufficiently fit to deploy to combat, might reduce the risk. The running number of observed person-months as of each record in the panel structure was encoded as a continuous variable in order to provide direct control for the passage of time, facilitating the discretetime regression analysis. This parameter accomplished a similar function as the time factor inherent to the semi-parametric Cox proportional hazards regression approach (Gibbons et al., 2003).
3. Results There were 396 incident back pain diagnoses among the 1248 subjects, representing a period prevalence of 31.7% for the 2134 person-years of observed time obtained by aggregating the available person-months. The back pain incidence rate was 18.6 cases per 100 person-years. Table 2 displays the frequencies and percentages of the last-observed characteristics among male and female subjects who did and did not experience incident back pain while observed. The results of Pearson’s chi square tests for differences in these frequencies are also provided. Results of tests of the unadjusted association between any prior depression and incident back pain were not statistically significant at the 95% confidence level. Among males, there were statistically-significant differences in the distributions of subjects by age, combat deployment experience and occupational area for the subpopulations with and without the outcome. There were no findings with statistical significance for occupational area among females, but such findings were revealed for race, age, prior service, tobacco use, and pregnancy. Table 3 demonstrates the adjusted odds ratios for each covariate computed from multivariable discrete-time logistic regression. Results have been reported as odds ratios rather than regression coefficients because logistic regression coefficients represent changes in log odds, which are much less interpretable than odds ratios. Among women, the presence of any past history of depression was associated with a 75% increase in the adjusted odds of back pain (95% CI: 1.08–2.84, Pvalue < 0.01) when compared to subjects without depression. While the effect size for depression was very similar among males, there was no statistical significance for this relationship in the smaller male subgroup. While the models revealed no statistically-significant findings for race, for both genders, the highest age group (37 years and older) was at the greatest adjusted risk for back pain. This finding stands in contrast to that of the two medium-age categories, for which there was no evidence of a difference in the back pain odds when compared to the youngest subjects. And, although no unadjusted effects were seen for marital status in the chi square analysis (Table 2), women who were single or married were at higher adjusted odds of back pain than those formerly married (Table 3). There were no results of statistical significance for prior military service in the multivariable analysis. However, the presence of combat deployments was associated with increases in the adjusted odds of back pain for each gender when compared to those without deployments.
2.3. Statistical analysis The panel data structure permitted a survival analysis in which subjects were initially observed upon entry to service as RNs. Observation ended upon the incident occurrence of the back pain 151
International Journal of Nursing Studies 74 (2017) 149–154
D.A. Nelson et al.
Table 2 Distributionsa of subjectb characteristics as of the last observation (N = 1248). Factor
Males with back pain 113 (29.2)
Males without back pain 274 (70.8)
Females with back pain 283 (32.9)
Females without back pain 578 (67.1)
History of depression Yes No
P = 0.572 6 (5.3) 107 (94.7)
11 (4.0) 263 (96.0)
20 (7.1) 263 (92.9)
39 (6.8) 539 (93.2)
Race White Black Asian/Pacific Islander
P = 0.854 83 (73.4) 10 (8.9) 20 (17.7)
199 (72.6) 21 (7.7) 54 (19.7)
P = 0.012 189 (66.8) 57 (20.1) 37 (13.1)
421 (72.8) 72 (12.5) 85 (14.7)
Age, years < 26 26–28 29–36 37 or greater
P = 0.034 27 (23.9) 18 (15.9) 37 (32.7) 31 (27.4)
71 75 80 48
P = 0.001 117 (41.3) 53 (18.7) 63 (22.3) 50 (17.7)
263 (45.5) 144 (24.9) 118 (20.4) 53 (9.2)
Marital status Married Never married Formerly married
P = 0.471 58 (51.3) 48 (42.5) 7 (6.2)
156 (56.9) 107 (39.0) 11 (4.0)
P = 0.560 133 (47.0) 135 (47.7) 15 (5.3)
251 (43.4) 290 (50.2) 37 (6.4)
Had prior military service Yes No
P = 0.065
Combat deployments None 1 2 or more
P = 0.010
Occupational area Medical/surgical Perioperative Obstetrics/ gynecology Psychiatric Critical care Emergency department Other
P = 0.049 86 (76.1) 4 (3.5) none
Body Mass Index, kg/m2 Normal/ underweight (24.99 or less) Overweight (25–29.99) Obese (30 or greater) No BMI data
31 (27.4) 82 (72.6)
70 (62.0) 34 (30.0) 9 (8.0)
P = 0.862
(25.9) (27.4) (29.2) (17.5)
52 (19.0) 222 (81.0)
78 (27.6) 205 (72.4)
510 (88.2) 52 (9.0) 16 (2.8)
194 (70.8) 1 (0.4) none
P = 0.143 236 (83.4) 8 (2.8) 2 (0.7)
452 (78.2) 15 (2.6) 18 (3.1)
2 (1.8) 17 (15.0) 2 (1.8)
7 (2.5) 49 (17.9) 19 (6.9)
1 (0.3) 22 (7.8) 5 (1.8)
2 (0.4) 66 (11.4) 14 (2.4)
2 (1.8)
4 (1.5)
9 (3.2)
11 (1.9)
P = 0.308
153 (54.0)
358 (61.9)
59 (52.2)
132 (48.2)
97 (34.3)
178 (30.8)
17 (15.0)
30 (11.0)
20 (7.1)
26 (4.5)
7 (6.2)
14 (5.0)
13 (4.6)
16 (2.8)
Tobacco user Yes No
P = 0.518 19 (16.8) 94 (83.2)
39 (14.2) 235 (85.8)
P = 0.017 18 (6.4) 265 (93.6)
17 (2.9) 561 (97.1)
Last 2 pregnancy trimesters Yes No
N/A
b
History of depression
1.72 (0.70–4.21)
1.75** (1.08–2.84)
Race White Black Other races
1.00 (referent) 0.74 (0.36–1.53) 0.92 (0.55–1.53)
1.00 (referent) 1.22 (0.89–1.68) 0.95 (0.66–1.37)
Age, years < 26 26–28 29–36 37 or greater
1.00 (referent) 1.21 (0.65–2.28) 1.46 (0.79–2.70) 2.25* (1.11–4.57)
1.00 (referent) 1.39 (0.98–1.98) 1.03 (0.69–1.52) 1.63* (1.05–2.55)
1.20 2.28 1.00 1.29
1.89* (1.09–3.28) 2.02* (1.12–3.63) 1.00 (referent) 1.25 (0.92–1.69)
Body Mass Index, kg/m Normal/underweight (24.99 or less) Overweight (25–29.99) Obese (30 or greater) No BMI data
P = 0.074 98 (35.8)
Odds ratios and 95% confidence intervals for females; n = 981 (69.9%)
Occupational area General medical/ surgical Perioperative Obstetrics/ gynecology Psychiatric Critical care Emergency department Other
P = 0.003 225 (79.5) 42 (14.8) 16 (5.7)
Odds ratios and 95% confidence intervals for males; n = 387 (30.1%)
Combat deployments None 1 2 or more
111 (19.2) 467 (80.8)
211 (77.0) 48 (17.5) 15 (5.5)
Factor
Marital status Married Never married Formerly married Had prior military service
P = 0.005
30 (35.8)
a
Table 3 Adjusted odds ratios, 95% confidence intervals and statistical significance indicatorsa from discrete-time multivariable logistic regression analyses (N = 1248).
Tobacco user Last 2 pregnancy trimesters Person-month of service a
—— ——
4 (1.4) 279 (98.6)
1.00 (referent) 1.83* (1.11–3.01) 2.00 (0.90–4.43)
1.00 (referent) 1.33 (0.90–1.95) 2.08* (1.16–3.72)
1.00 (referent)
1.00 (referent)
9.88*** (3.21–30.4) N/A
1.76 (0.84–3.69) 0.77 (0.18–3.18)
1.60 (0.34–7.45) 1.25 (0.71–2.20) 0.52 (0.12–2.18)
0.83 (0.11–6.17) 0.98 (0.62–1.57) 1.08 (0.43–2.70)
1.20 (0.27–5.38)
1.38 (0.68–2.82)
1.00 (referent)
1.00 (referent)
1.34 (0.84–2.12)
1.11 (0.84–1.45)
2.03* (1.08–3.82) 0.37* (0.15–0.87)
2.07** (1.24–3.47) 0.36** (0.20–0.64)
1.44 (0.85–2.44) N/A
1.50 (0.91–2.47) 0.49 (0.18–1.35)
0.97*** (0.95–0.99)
0.98*** (0.96–0.99)
2
Statistical significance: *** < 0.001; ** < 0.01; * < 0.05.
resulting from the low numbers of subjects in this category. Obese BMI was associated with a very similar doubling of the adjusted back pain odds for men and women. However, there was no evidence that subjects in the overweight BMI range differed in these odds from those in the normal range. For both genders, there was a large decrease in the odds of back pain when subjects had no known BMI data. As mentioned, this effect was potentially due to the fact that much of the BMI data in this population arose from health care. Individuals needing little or no care might be the most resistant to back pain in this environment, and would also demonstrate minimal BMI data. We did not discover statistical significance for tobacco use for either gender in the multivariable analysis, nor for pregnancy among the female subjects. In both models, each additional person-month of observation was associated with small but statistically-significant decreases in the adjusted odds of back pain.
P = 0.024 —— ——
(0.51–2.81) (0.89–5.84) (referent) (0.80–2.08)
1 (0.2) 577 (99.8)
Statistical significance levels for comparisons were computed using chi square tests. Format: frequency (%).
Among men, there was an 83% increase in these odds when the subject had one deployment (95% confidence interval [CI]: 1.11–3.01). The back pain odds doubled among women with two or more deployments (OR: 2.08; 95% CI: 1.16–3.72). The only finding of statistical significance for occupational area was for perioperative work among males. It was associated with a 9.9-fold increase in the adjusted odds of back pain, but with a wide confidence interval (95% CI: 3.22–30.3) 152
International Journal of Nursing Studies 74 (2017) 149–154
D.A. Nelson et al.
4. Discussion
substantial back pain and depression events that preceded the observed time period, despite our careful examination of total data on the subjects. It is therefore possible that for some subjects, undetected diagnoses of these problems and temporal relationships between them existed. It does not appear possible to estimate the magnitude of this effect, but such information could have biased our results towards the null hypothesis, strengthening our findings. We suggest that the Army's requirements that no serious depression or back pain problems are present among newly-commissioned officers reduce the concern (US Army, 2011). The study was also limited by the inability to assume the findings will extend to the general population or to civilian nurses. We further recognize that the multiple data sources available to the study, such as military-specific sources of BMI data and diagnosis codes for back pain and depression, may not have undergone prior validation in the extant literature. Further research using techniques such as out-of-sample validation would be needed to fully validate our findings. Also, we employed binary indicators for depression and back pain. Specific severity levels of these problems could benefit from further study in order to clarify more subtle relationships that may be present. Additionally, we recognize the potential roles of other factors as back pain predictors, such as anxiety and post-traumatic stress disorder. These factors were not part of this study, but we expect to examine them in ongoing research. Finally, we used retrospective rather than prospective data for this study under an observational approach, methods that can be limited by matters including the absence of a control group. However, the use of these data has provided benefits including the ability to observe natural phenomena in the total target population without bias that might be introduced in experimental designs, such as the Hawthorne effect.
While depression has been associated with back pain (Hoy et al., 2010; Nicholas et al., 2011), the predominant direction of causality has been uncertain. The main finding of this study was a statistically-significant increase in the adjusted odds of later, incident back pain among female RNs when prior depression was present. The extremely-similar effect estimate for the male subpopulation lacked statistical significance, but this relationship deserves further study due to the small male RN group and their attendant low number of outcomes. Another important finding was that the adjusted back pain odds for males and females were approximately doubled when BMI was 30 or greater. The findings help to clarify the impact of BMI on back pain. BMI in the overweight range was associated with back pain in U.S. nurses in one cross-sectional study (Darby et al., 2013), but a casecontrol study of Swedish nurses found no association between back injuries and BMI (Engkvist et al., 2000). Our research supports the former finding. Overall, our findings suggest that careful management of depression and body composition may pay dividends in reducing back pain rates in the nursing profession. The covariates selected for this multivariable analysis were also important in order to adequately address potential confounding. Depression is associated with elevated BMI (Dong et al., 2004), and as noted above, elevated BMI was known to be a potential back pain predictor at the outset of this work. We believe our study is the first to shed light on the potentially-complex temporal relationship between a history of depression and back pain when taking BMI into account, the first to do so using comprehensive, multisource longitudinal data, and the first to do so in the nursing profession. The rates of back pain we discovered in this population were similar to rates reported in literature on the civilian workforce, where U.S. nurses were found to have a 12-month prevalence of 32% (Trinkoff et al., 2003) and a 15-month back pain incidence of 21.1% (Trinkoff et al., 2006). We recognize that these comparisons may be limited by our definition for back pain, which was mostly based on provider diagnoses, whereas the cited studies of civilian nurses used self-report questionnaires. However, the comparisons to civilian rates appear reasonable, as we did include self-report from Periodic Health Assessment data, and overall, there is no definitive, objective test for the presence of back pain. The evaluation of back pain is not well-standardized (Dagenais et al., 2010), and all providers must substantially rely on patient self-report when making their diagnoses. The similarities between prior civilian nurse back pain rates and our findings may be due to relatively few differences between main workplace exposures for civilian and military nurses. While both groups work nominal 40-h work weeks, per communication with the office of the Chief of the US Army Nurse Corps, October 2016, working hours among Army nurses can vary widely along with exposure to manual patient handling training. Similarly, many civilian hospitals have mandatory overtime policies, requiring nurses to stay longer than their usual 12 h shifts. In the absence of specific ergonomic regulations from the Occupational Safety and Health Administration, neither the overall civilian hospital system nor military hospitals (MAMC, 2014) are known to have standardized low lift policies or consistent safe patient handling programs. Combat duty is one unique exposure for the military and its nurses, and it was associated with increases in the adjusted odds of back pain among the RNs we studied. The Army's physical fitness policies and practices (US Army, 2012) that mandate regular exercise and testing for physical performance represent another difference from the civilian workplace, which has no such requirements. However, we find no evidence that the military fitness program itself creates an increased risk for back pain. Whether perioperative work creates increased back pain risk deserves further study, as our sample was sufficiently small to suggest that the finding could have arisen from chance. Other limitations of this work include our inability to rule out
5. Conclusion In conclusion, this study has produced new information on the epidemiology and predictors of back pain among military nurses. Our findings provide actionable occupational health information that might assist military health system leaders in addressing the behavioral health needs of their nurses. The confirmed impact of high BMI on back pain that we found suggests that improved BMI may reduce back pain risk. Studies should be conducted on the impact on back pain rates of military programs such as the Army’s Performance Triad (US Army, 2015b), which has been conceived to help soldiers improve their health, including by reducing BMI. In addition, the Army and civilian health systems might consider standardizing the training of medical personnel in patient handling and the equipment used across facilities in order to reduce overall back pain rates. Such a policy change could provide a positive impact, given that the manual handling of patients is a known back pain risk factor for both military and civilian nurses. Funding This research was supported by an unnumbered seed grant from Leidos, Reston, VA to the University of Maryland. Funders played no role in study design, analysis or the reporting of results. Acknowledgments The authors acknowledge the contributions of the following organizations to our ability to successfully conduct this project: the US Army Office of the Surgeon General, and the Center for Health Informatics and Decision Systems of the Robert H. Smith School of Business, the University of Maryland, College Park. References CDC: Centers for Disease Control and Prevention, 2015. About Adult BMI. http://www.
153
International Journal of Nursing Studies 74 (2017) 149–154
D.A. Nelson et al.
Nelson, D.A., Wolcott, V.L., Kurina, L.M., 2016. Prediction of all-cause occupational disability among US Army soldiers. Occup. Environ. Med. 0, 1–10. http://dx.doi.org/ 10.1136/oemed-2015-103436. Nicholas, M.K., Linton, S.J., Watson, P.J., Main, C.J., 2011. Early identification and management of psychological risk factors (yellow flags) in patients with low back pain: a reappraisal. Phys. Ther. 91 (5), 1–17. Niebuhr, D.W., Krampf, R.L., Mayo, J.A., et al., 2011. Risk factors for disability retirement among healthy adults joining the U.S. Army Mil. Med. 176, 170–175. Patzkowski, J.C., Rivera, J.C., Ficke, J.R., Wenker, J.C., 2012. The changing face of disability in the US army: the operation enduring freedom and operation iraqi freedom effect. J. Am. Acad. Orthop. Surg. 20 (Suppl. 1), 23–30. Pinheiro, M.B., Ferreira, M.L., Refshauge, K., et al., 2015. Symptoms of depression and risk of new episodes of low back pain: a systematic review and meta-analysis. Arthritis Care Res. 67 (11), 1591–1603. Pinheiro, M.B., Ferreira, M.L., Refshauge, K., et al., 2016. Symptoms of depression as a prognostic factor for low back pain: a systematic review. Spine J. 16 (1), 105–116. Rosellini, A.J., et al., 2016. Predicting non-familial major physical violent crime perpetration in the US Army from administrative data. Psychol. Med. 46, 303–316. Shiri, R., Karppinen, J., Leino-Arjas, P., et al., 2010a. The association between obesity and low back pain: a meta-analysis. Am. J. Epidemiol. 171, 135–154. Shiri, R., Karppinen, J., Leino-Arjas, P., Solovieva, S., Viikari-Juntura, E., 2010b. The association between smoking and low back pain: a meta-analysis. Am. J. Med. 123, 87 e7-35. Sikorski, C., Emerson, M.A., Cowan, D.N., et al., 2012. Risk factors for medical disability in U.S. enlisted Marines: fiscal years 2001 to 2009. Mil. Med. 177, 128–134. Trinkoff, A.M., Lipscomb, J.A., Geiger-Brown, J., Storr, C.L., Brady, B.A., 2003. Perceived physical demands and reported musculoskeletal problems in registered nurses. Am. J. Prev. Med. 24 (3), 270–275. Trinkoff, A.M., Le, R., Geiger-Brown, J., Lipscomb, J.A., Lang, G., 2006. Longitudinal relationship of work hours, mandatory overtime, and on-call to musculoskeletal problems in nurses. Am. J. Ind. Med. 49 (11), 964–971. US Army, 2011. Army Regulation 40-501: Standards of Medical Fitness. http://www.apd. army.mil/pdffiles/r40_501.pdf. (Accessed 17 April 2012). US Army, 2012. Field Manual 7-22: Army Physical Readiness Training. http://www.apd. army.mil/pdffiles/fm7_22.pdf. (Accessed 17 April 2012). US Army, 2013. Army Regulation 600-9: The Army Body Composition Program. http:// www.apd.army.mil/pdffiles/ar600_9.pdf. (Accessed 17 April 2012). US Army, 2015a. Medical Operational Data System. https://g1arng.army.pentagon.mil/ Systems/MODS/Pages/default.aspx. (Accessed 17 April 2012). US Army, 2015b. How the Army’s Performance Triad Can Help You Reach Your Healthy Living Goals. http://www.health.mil/News/Articles/2015/03/03/How-the-ArmysPerformance-Triad-Can-Help-You-Reach-Your-Healthy-Living-Goals?type=Articles& page=5#pagingAnchor. (Accessed 16 November 2017). Vermani, E., Mittal, R., Weeks, A., 2010. Pelvic girdle pain and low back pain in pregnancy: a review. Pain Pract. 10 (1), 60–71. Yassi, A., Lockhart, K., 2013. Work-relatedness of low back pain in nursing personnel: a systematic review. Int. J. Occup. Environ. Health 19 (3), 223–244.
cdc.gov/healthyweight/assessing/bmi/adult_bmi/. (Accessed 28 July 2016). Cohen, S.P., Nguyen, C., Kapoor, S.G., et al., 2009. Back pain during war: an analysis of factors affecting outcomes. Arch. Intern. Med. 169, 1916–1923. Cohen, S.P., Brown, C., Kurihara, C., Plunkett, A., Hguyen, C., Strassels, S.A., 2010. Diagnoses and factors associated with medical evacuation and return to duty for service members participating in Operation Iraqi Freedom or Operation Enduring Freedom: a prospective cohort study. Lancet 375, 301–309. Dagenais, S., Tricco, A.C., Halderman, S., 2010. Synthesis of recommendations for the assessment and management of low back pain from recent clinical practice guidelines. Spine J. 10 (6), 514–529. Darby, B., Gallo, A.M., Fields, W., 2013. Physical attributes of endoscopy nurses related to musculoskeletal problems. Gastroenterol. Nurs. 36 (3), 202–208. Davis, K.G., Kotowski, S.E., 2015. Prevalence of musculoskeletal disorders for nurses in hospitals, long-term care facilities, and home health care: a comprehensive review. Hum. Factors 57 (5), 754–792. de Schepper, E.I., et al., 2010. The association between lumbar disc degeneration and low back pain: the influence of age, gender and individual radiographic features. Spine 35 (5), 531–536. Dong, C., Sanchez, L.E., Price, R.A., 2004. Relationship of obesity to depression: a familybased study. Int. J. Obes. 28, 790–795. Engkvist, I.L., Hjelm, E.W., Hagberg, M., Menckel, E., Ekenvall, L., 2000. Risk indicators for reported over-exertion back injuries among female nursing personnel. Epidemiology 11 (5), 519–522. Gaderman, A.M., Engel, C.C., Naifeh, J.A., Nock, M.K., Petkhova, M., Santiago, P.N., Benjamin, W., Zaslavsky, A.M., Kessler, R.C., 2012. Prevalence of DSM-IV major depression among US military personnel: meta-analysis and simulation. Mil. Med. 177 (8), 47–59. Gibbons, R.D., et al., 2003. Waiting for organ transplantation: results of an analysis by an Institute of Medicine Committee. Biostatistics 4 (2), 207–222. Hoge, C.W., Castro, C.A., Messer, S.C., McGurk, D., Cotting, D.I., Koffman, R.L., 2004. Combat duty in Iraq and Afghanistan, mental health problems, and barriers to care. N. Engl. J. Med. 351 (1), 13–22. Hoy, D., Brooks, P., Blyth, F., Buchbinder, R., 2010. The epidemiology of low back pain. Best Pract. Res. Clin. Rheumatol. 24 (6), 769–781. Knox, J., Orchowski, J., Scher, D.L., Owens, B.D., Burks, R., Belmont, P.J., 2011. The incidence of low back pain in active duty United States military service members. Spine 36 (18), 1492–1500. Letvak, S., Ruhm, C.J., McCoy, T., 2012. Depression in hospital-employed nurses. Clin. Nurse Spec. 26 (3), 177–182. Lincoln, A.E., Smith, G.S., Amoroso, P.J., Bell, N.S., 2002. The natural history and risk factors of musculoskeletal conditions resulting in disability among US Army personnel. Work 18, 99–113. MAMC: Madigan Army Medical Center, 2014. MAMC Memorandum Number 385-9. 13 May. Available upon request. Nelson, D.A., Kurina, L.M., 2013. Clinical prediction of musculoskeletal-related ‘Medically Not Ready’ for combat duty statuses among active duty U. S. Army soldiers. Mil. Med. 178, 1365–1372.
154
Contents lists available at ScienceDirect
International Journal of Nursing Studies journal homepage: www.elsevier.com/locate/ijns
Prior depression and incident back pain among military registered nurses: A retrospective cohort study
MARK
⁎
D. Alan Nelsona, , Nancy Menzelb, Patricia Horohoc a b c
Department of Medicine, Stanford University School of Medicine, 450 Serra Mall, Bldg 20, Stanford, CA, 94305-2160, USA University of Nevada, Las Vegas, School of Nursing, 4505 South Maryland Parkway, Box 453018, Las Vegas, NV, 89154-3018, USA 800 North Glebe Road, Suite 300, Arlington, VA, 22203-1807, USA
A R T I C L E I N F O
A B S T R A C T
Keywords: Back ache Lower back pain Depression Epidemiology Registered nurses Occupational health Patient handling
Background: Occupational back pain rates are substantial among registered nurses, and nurses also report high rates of depression. The role of depression as a potential predictor of back pain among nurses appears understudied. Objectives: The objective of the study was to determine whether a history of depression predicted incident back pain in a population of military registered nurses when controlling for relevant risk factors. Design: We employed a retrospective cohort approach using longitudinal data in which gender-specific subject groups were followed from the beginning of duty as a registered nurse to the occurrence of an outcome, or to censoring due to completion of service or the end of available data. Participants: This study included all United States Army registered nurses who began work during 2011–2014 without evidence of prior back pain in clinical records. Methods: Data from automatically-collected medical and administrative sources were combined and used to provide 2134 person-years of observation on 1248 individuals. These data were organized at the person-month level in a panel data structure to support discrete-time multivariable logistic regression models. The models examined the relationships between prior depression, Body Mass Index, the presence of prior combat duty and selected control variables and the outcome, the incident occurrence of back pain. Results: The incidence rate of back pain was 18.6 per 100 person-years and the period prevalence was 31.7%. Prior depression was a statistically-significant predictor of incident back pain among female subjects (odds ratio [OR]: 1.75, 95% confidence interval [CI]: 1.08–2.83, P-value < 0.05). Body Mass Index of 30 kg/m2 or greater, prior combat deployments, and age 36 years or older was each associated with back pain for male and female nurses. Conclusions: The study’s findings provide the first evidence of a temporal link between antecedent depression and later back pain among female military nurses. High Body Mass Index was found to be a further, modifiable risk factor for back pain in this population.
What this paper adds
What is already known about the topic?
• Depression and back pain rates are relatively high among registered nurses and military service members. • Back pain and depression have long been associated in the litera-
ture, but there has been little work on the possible temporal relationships between these problems when controlling for other, potential contributory factors.
⁎
Corresponding author. E-mail address: [email protected] (D.A. Nelson).
http://dx.doi.org/10.1016/j.ijnurstu.2017.06.015 Received 16 December 2016; Received in revised form 22 June 2017; Accepted 23 June 2017 0020-7489/ © 2017 Elsevier Ltd. All rights reserved.
• Prior depression was associated with similar increases in the ad•
justed odds of back pain for both genders, but this association was statistically-significant among female nurses only, probably due to reduced power for males. Body Mass Index of 30 kg/m2 was associated with a similar, greater than two-fold increase in the adjusted odds of back pain among the 1248 male and female nurses studied.
International Journal of Nursing Studies 74 (2017) 149–154
D.A. Nelson et al.
1. Introduction
Table 1 Data sources leveraged to produce the panel dataset for analysis.
Registered nurses (RNs) report high rates of occupational back pain. A review of 132 international studies of workplace problems among civilian nurses (Davis and Kotowski, 2015) found that the mean, pastyear back pain among nurses working in hospitals (66 studies) was 55%. While the RN workforce faces unique health risk factors and problems, the predictors of back pain among RNs have not been fully explored. Requirements such as patient handling are known to contribute to back pain in the nursing profession (Yassi and Lockhart, 2013). However, the literature reveals little work on the relationships between mental health statuses and back pain among RNs. This deficit is noteworthy because depression rates are about twice as high among RNs working in hospitals as in the general population (Letvak et al., 2012). Depression is a common condition that is associated with back pain across multiple studies in the research literature (Hoy et al., 2010; Nicholas et al., 2011; Pinheiro et al., 2015). Depression has also been associated with a poorer prognosis in the course of back pain (Pinheiro et al., 2016). Altogether, these ideas suggest that new research addressing the relationship between depression and back pain is sorely needed in the RN population. The military provides one structured environment in which to explore the depression-back pain association. In the total United States (U.S.) Army population, back pain is a leading problem associated with lost work time (Knox et al., 2011), medical evacuation during operational deployment (Cohen et al., 2009, 2010) and disability (Lincoln et al., 2002; Patzkowski et al., 2012). Depression is also prevalent in the U.S. military (Gaderman et al., 2012; Hoge et al., 2004). However, as is also applicable to civilian RNs, we find scant past research that meaningfully addresses back pain or depression among military RNs. To address the gaps in the literature that we identified, we designed this study to examine incident back pain diagnoses among U.S. Army commissioned officers who began duty as RNs during 2011–2014. The specific aim was to assess whether a history of prior depression predicted later back pain when studying longitudinal data and controlling for other, potential back pain predictors.
Data source
Description
Defense Manpower Data Center (DMDC) Active Duty Master file Transactions file
Demographics and military service data Discharges from service
Military Health System Data Repository (MDR) Combined Ambulatory Outpatient care in military facilities Professional Encounter Record Clinical Data Repository Vital Height and weight readings from Signs outpatient encounters Standard Inpatient Data Record Inpatient care in military facilities Tricare Encounter Data, NonOutpatient care in civilian facilities Institutional Tricare Encounter Data, Inpatient care in civilian facilities Institutional Defense Training Management System (DTMS) Height/weight file Height and weight readings from biannual body composition checks Medical Operations Data System (MODS) (US Army, 2015a) Periodic Health Assessment Height and weight readings and selfreported medical problems eProfile Duty restrictions assigned by clinicians
were taken during data cleaning at the US Army Office of the Surgeon General to mitigate or eliminate missing data through cross-referencing of data elements across multiple sources before the final, de-identified dataset was provided to the University of Maryland. The sole, identifiable missing data were for Body Mass Index, addressed below. 2.2.1. Dependent variable As observed by Pinheiro et al. (2015) in a systematic review and meta-analysis of studies addressing back pain and depression, back pain definitions that result from health care-seeking or reduced functional capacity are the norm in many studies. We employed similar conceptual bases for the definition of back pain in the study population. The dichotomous dependent variable was assigned the “1” value in the person-month in which we first observed documentation of low back pain or other back ache in clinical encounter data. The International Classification of Disease, Ninth Revision, Clinical Modification (ICD-9-CM) system remained the diagnostic system for the military as of the time of the observed data. Eligible clinical encounters or hospital admissions were those in which selected ICD-9-CM codes documented lumbar, thoracic, or regionally-unspecified back pain (724.1, 724.2, 724.5, 847.1, 847.2 or 847.9). A “1” value for the outcome variable was also assigned when back pain was entered as a clinical condition in an assigned duty restriction reflecting short- or long-term disability. Finally, back pain was defined as being present if the patient reported back pain by selecting a formatted field for this condition in a Periodic Health Assessment screening document.
2. Methods 2.1. Population and dataset The study population was established using a comprehensive, deidentified medical and administrative database on the total U.S. Army at the Center for Health Information and Decision Systems of the University of Maryland, College Park. Service members from other U.S. military branches such as the U.S. Navy were not available for study. In order to assess incident back pain after subjects began work as military RNs, our potentially-eligible subjects for the study were 1320 individuals who newly entered service as such during 2011–2014. We required that there was no evidence of prior back pain in available data from prior care and duty restrictions, rendering 72 potential subjects ineligible to produce the study population of 1248 individuals. The analytic dataset was a longitudinal panel that included military service, administrative and demographic data on the eligible subjects. The information was organized at the person-month level, with timevarying factors arranged in temporal sequence for each subject. The data sources used to produce the dataset are listed in Table 1.
2.2.2. Independent variables Increasing age may be associated with back pain (Hoy et al., 2010; Knox et al., 2011) and radiographically-verifiable spinal degeneration (de Schepper et al., 2010). Each subject's running age in years was classified using a categorical variable based on frequency distributions. Age was categorized rather than expressed as a continuous value due to non-linear effects, to be reported. The age categories were < 26 years (reference group), 26–28, 28–36, and > 36. The models also controlled for the subject’s self-reported race due to prior evidence of differences in disability rates for race-based groups (Niebuhr et al., 2011; Sikorski et al., 2012). We included a categorical variable for marital status in light of the previous association between being married and back pain in military populations (Knox et al., 2011). Subjects were classified as married, never married, or formerly married (divorced, separated or widowed).
2.2. Variables Data were obtained from official sources upon which the Department of Defense depends for critical operations. Given its ability to enforce data collection policies for service members, the demographic and most of the clinically-related variables employed in this study were remarkably free of known missing values. Additional efforts 150
International Journal of Nursing Studies 74 (2017) 149–154
D.A. Nelson et al.
outcome, departure from active military service, or the end of the available data in December 2014. When totaling the available personmonths of service time, the final dataset provided 2134 observed person-years on the 1248 subjects (mean: 20.5 months; SD: 13.3 months; range: 1–48 months.) Prior to regression analyses, we computed chi square tests for distribution differences among covariate categories for subjects with and without outcomes in order to detect potentially-useful associations. Stratified, discrete-time multivariable logistic regression models for males and females were then computed due to the frequent, widespread effect modification by gender seen in prior research on a range of health and other outcomes in the military (Nelson et al., 2016; Nelson and Kurina, 2013; Rosellini et al., 2016). Numerous, preliminary models were explored using the range of potential variable configurations that applied to each factor of interest. Final variable configurations were attained that provided maximum utility and realization of associations. This research study was approved by the Institutional Review Board at the University of Maryland and underwent secondary review by the Defense Health Agency’s Human Research Protection Office. The research has conformed to the principles embodied in the Declaration of Helsinki. All statistical analyses were conducted using Stata 14 software (StataCorp, College Station, Texas).
Body Mass Index (BMI) is an important factor due to the prior association of excess weight and back pain (Shiri et al., 2010a). We incorporated BMI as a categorical covariate due to non-linear effects to be reported. Our BMI data originated from height and weight readings taken at elective outpatient encounters, annual health assessments, and biannual administrative readings (US Army, 2013). We included a category for missing BMI data due to the presence of subjects with few or no health encounters at which readings might be taken. This was an important approach because low care frequency may correlate with less back pain. Aside from the category for missing data, the variable was organized based on the standard BMI categories (CDC, 2015). “Underweight” and “normal” ranges were combined in the reference category due to rarity of low BMI in this population. We also provided control for tobacco use due to its association with back pain (Shiri et al., 2010b). This dichotomous covariate indicated any prior self-report of tobacco use at outpatient health encounters or annual assessments. To examine the temporal relationship between prior depression and back pain in our study population, we included a rolling, dichotomous covariate indicating whether any observed evidence of depression preceded each observation. Depression events were out- or inpatient clinical diagnoses using selected ICD-9-CM codes (311, 296.2, 296.3, 309.0 or 309.1). Applicable events were also defined using documentation of back pain in Periodic Health Assessments or duty restrictions. A further binary covariate was assigned a “1” value in the longitudinal data during the last two trimesters of pregnancy, where applicable. This status was identified using required documentation of pregnancy for female soldiers in duty restrictions found in ancillary health files. We included this variable in order to control for pregnancyassociated back pain, which is a common problem for pregnant women (Vermani et al., 2010). This covariate also provided control for the military's regulation-mandated reductions in job duties during advanced pregnancy (US Army, 2011), which could conceivably reduce the risk of back pain through reduced workloads. The variable was further intended to reveal any confounding of the concurrent BMI variable, given the expected increases in BMI during later pregnancy. To control for potential variations in job duties, each nurse’s specialty area was encoded as a categorical variable. These classifications were based on the Defense Manpower Data Center records of each nurse's official primary Area of Concentration, the military term for commissioned officer job type, and additional skill identifiers. Because some Army nurses served in the military in other capacities prior to being commissioned as officers for nursing duty, such as in enlisted service or as another officer type, we identified any prior military service with a binary covariate. We also incorporated a categorical variable for the number (if any) of combat deployment experiences preceding each person-month in order to provide control for the potential effects associated with this form of military duty. This variable was included because it appeared unclear whether the mental and physical stress associated with deployments could increase the risk of back pain. An opposing theory could hold that a “healthy warrior” selection effect seen in other research (Niebuhr et al., 2011), associated with having previously been found sufficiently fit to deploy to combat, might reduce the risk. The running number of observed person-months as of each record in the panel structure was encoded as a continuous variable in order to provide direct control for the passage of time, facilitating the discretetime regression analysis. This parameter accomplished a similar function as the time factor inherent to the semi-parametric Cox proportional hazards regression approach (Gibbons et al., 2003).
3. Results There were 396 incident back pain diagnoses among the 1248 subjects, representing a period prevalence of 31.7% for the 2134 person-years of observed time obtained by aggregating the available person-months. The back pain incidence rate was 18.6 cases per 100 person-years. Table 2 displays the frequencies and percentages of the last-observed characteristics among male and female subjects who did and did not experience incident back pain while observed. The results of Pearson’s chi square tests for differences in these frequencies are also provided. Results of tests of the unadjusted association between any prior depression and incident back pain were not statistically significant at the 95% confidence level. Among males, there were statistically-significant differences in the distributions of subjects by age, combat deployment experience and occupational area for the subpopulations with and without the outcome. There were no findings with statistical significance for occupational area among females, but such findings were revealed for race, age, prior service, tobacco use, and pregnancy. Table 3 demonstrates the adjusted odds ratios for each covariate computed from multivariable discrete-time logistic regression. Results have been reported as odds ratios rather than regression coefficients because logistic regression coefficients represent changes in log odds, which are much less interpretable than odds ratios. Among women, the presence of any past history of depression was associated with a 75% increase in the adjusted odds of back pain (95% CI: 1.08–2.84, Pvalue < 0.01) when compared to subjects without depression. While the effect size for depression was very similar among males, there was no statistical significance for this relationship in the smaller male subgroup. While the models revealed no statistically-significant findings for race, for both genders, the highest age group (37 years and older) was at the greatest adjusted risk for back pain. This finding stands in contrast to that of the two medium-age categories, for which there was no evidence of a difference in the back pain odds when compared to the youngest subjects. And, although no unadjusted effects were seen for marital status in the chi square analysis (Table 2), women who were single or married were at higher adjusted odds of back pain than those formerly married (Table 3). There were no results of statistical significance for prior military service in the multivariable analysis. However, the presence of combat deployments was associated with increases in the adjusted odds of back pain for each gender when compared to those without deployments.
2.3. Statistical analysis The panel data structure permitted a survival analysis in which subjects were initially observed upon entry to service as RNs. Observation ended upon the incident occurrence of the back pain 151
International Journal of Nursing Studies 74 (2017) 149–154
D.A. Nelson et al.
Table 2 Distributionsa of subjectb characteristics as of the last observation (N = 1248). Factor
Males with back pain 113 (29.2)
Males without back pain 274 (70.8)
Females with back pain 283 (32.9)
Females without back pain 578 (67.1)
History of depression Yes No
P = 0.572 6 (5.3) 107 (94.7)
11 (4.0) 263 (96.0)
20 (7.1) 263 (92.9)
39 (6.8) 539 (93.2)
Race White Black Asian/Pacific Islander
P = 0.854 83 (73.4) 10 (8.9) 20 (17.7)
199 (72.6) 21 (7.7) 54 (19.7)
P = 0.012 189 (66.8) 57 (20.1) 37 (13.1)
421 (72.8) 72 (12.5) 85 (14.7)
Age, years < 26 26–28 29–36 37 or greater
P = 0.034 27 (23.9) 18 (15.9) 37 (32.7) 31 (27.4)
71 75 80 48
P = 0.001 117 (41.3) 53 (18.7) 63 (22.3) 50 (17.7)
263 (45.5) 144 (24.9) 118 (20.4) 53 (9.2)
Marital status Married Never married Formerly married
P = 0.471 58 (51.3) 48 (42.5) 7 (6.2)
156 (56.9) 107 (39.0) 11 (4.0)
P = 0.560 133 (47.0) 135 (47.7) 15 (5.3)
251 (43.4) 290 (50.2) 37 (6.4)
Had prior military service Yes No
P = 0.065
Combat deployments None 1 2 or more
P = 0.010
Occupational area Medical/surgical Perioperative Obstetrics/ gynecology Psychiatric Critical care Emergency department Other
P = 0.049 86 (76.1) 4 (3.5) none
Body Mass Index, kg/m2 Normal/ underweight (24.99 or less) Overweight (25–29.99) Obese (30 or greater) No BMI data
31 (27.4) 82 (72.6)
70 (62.0) 34 (30.0) 9 (8.0)
P = 0.862
(25.9) (27.4) (29.2) (17.5)
52 (19.0) 222 (81.0)
78 (27.6) 205 (72.4)
510 (88.2) 52 (9.0) 16 (2.8)
194 (70.8) 1 (0.4) none
P = 0.143 236 (83.4) 8 (2.8) 2 (0.7)
452 (78.2) 15 (2.6) 18 (3.1)
2 (1.8) 17 (15.0) 2 (1.8)
7 (2.5) 49 (17.9) 19 (6.9)
1 (0.3) 22 (7.8) 5 (1.8)
2 (0.4) 66 (11.4) 14 (2.4)
2 (1.8)
4 (1.5)
9 (3.2)
11 (1.9)
P = 0.308
153 (54.0)
358 (61.9)
59 (52.2)
132 (48.2)
97 (34.3)
178 (30.8)
17 (15.0)
30 (11.0)
20 (7.1)
26 (4.5)
7 (6.2)
14 (5.0)
13 (4.6)
16 (2.8)
Tobacco user Yes No
P = 0.518 19 (16.8) 94 (83.2)
39 (14.2) 235 (85.8)
P = 0.017 18 (6.4) 265 (93.6)
17 (2.9) 561 (97.1)
Last 2 pregnancy trimesters Yes No
N/A
b
History of depression
1.72 (0.70–4.21)
1.75** (1.08–2.84)
Race White Black Other races
1.00 (referent) 0.74 (0.36–1.53) 0.92 (0.55–1.53)
1.00 (referent) 1.22 (0.89–1.68) 0.95 (0.66–1.37)
Age, years < 26 26–28 29–36 37 or greater
1.00 (referent) 1.21 (0.65–2.28) 1.46 (0.79–2.70) 2.25* (1.11–4.57)
1.00 (referent) 1.39 (0.98–1.98) 1.03 (0.69–1.52) 1.63* (1.05–2.55)
1.20 2.28 1.00 1.29
1.89* (1.09–3.28) 2.02* (1.12–3.63) 1.00 (referent) 1.25 (0.92–1.69)
Body Mass Index, kg/m Normal/underweight (24.99 or less) Overweight (25–29.99) Obese (30 or greater) No BMI data
P = 0.074 98 (35.8)
Odds ratios and 95% confidence intervals for females; n = 981 (69.9%)
Occupational area General medical/ surgical Perioperative Obstetrics/ gynecology Psychiatric Critical care Emergency department Other
P = 0.003 225 (79.5) 42 (14.8) 16 (5.7)
Odds ratios and 95% confidence intervals for males; n = 387 (30.1%)
Combat deployments None 1 2 or more
111 (19.2) 467 (80.8)
211 (77.0) 48 (17.5) 15 (5.5)
Factor
Marital status Married Never married Formerly married Had prior military service
P = 0.005
30 (35.8)
a
Table 3 Adjusted odds ratios, 95% confidence intervals and statistical significance indicatorsa from discrete-time multivariable logistic regression analyses (N = 1248).
Tobacco user Last 2 pregnancy trimesters Person-month of service a
—— ——
4 (1.4) 279 (98.6)
1.00 (referent) 1.83* (1.11–3.01) 2.00 (0.90–4.43)
1.00 (referent) 1.33 (0.90–1.95) 2.08* (1.16–3.72)
1.00 (referent)
1.00 (referent)
9.88*** (3.21–30.4) N/A
1.76 (0.84–3.69) 0.77 (0.18–3.18)
1.60 (0.34–7.45) 1.25 (0.71–2.20) 0.52 (0.12–2.18)
0.83 (0.11–6.17) 0.98 (0.62–1.57) 1.08 (0.43–2.70)
1.20 (0.27–5.38)
1.38 (0.68–2.82)
1.00 (referent)
1.00 (referent)
1.34 (0.84–2.12)
1.11 (0.84–1.45)
2.03* (1.08–3.82) 0.37* (0.15–0.87)
2.07** (1.24–3.47) 0.36** (0.20–0.64)
1.44 (0.85–2.44) N/A
1.50 (0.91–2.47) 0.49 (0.18–1.35)
0.97*** (0.95–0.99)
0.98*** (0.96–0.99)
2
Statistical significance: *** < 0.001; ** < 0.01; * < 0.05.
resulting from the low numbers of subjects in this category. Obese BMI was associated with a very similar doubling of the adjusted back pain odds for men and women. However, there was no evidence that subjects in the overweight BMI range differed in these odds from those in the normal range. For both genders, there was a large decrease in the odds of back pain when subjects had no known BMI data. As mentioned, this effect was potentially due to the fact that much of the BMI data in this population arose from health care. Individuals needing little or no care might be the most resistant to back pain in this environment, and would also demonstrate minimal BMI data. We did not discover statistical significance for tobacco use for either gender in the multivariable analysis, nor for pregnancy among the female subjects. In both models, each additional person-month of observation was associated with small but statistically-significant decreases in the adjusted odds of back pain.
P = 0.024 —— ——
(0.51–2.81) (0.89–5.84) (referent) (0.80–2.08)
1 (0.2) 577 (99.8)
Statistical significance levels for comparisons were computed using chi square tests. Format: frequency (%).
Among men, there was an 83% increase in these odds when the subject had one deployment (95% confidence interval [CI]: 1.11–3.01). The back pain odds doubled among women with two or more deployments (OR: 2.08; 95% CI: 1.16–3.72). The only finding of statistical significance for occupational area was for perioperative work among males. It was associated with a 9.9-fold increase in the adjusted odds of back pain, but with a wide confidence interval (95% CI: 3.22–30.3) 152
International Journal of Nursing Studies 74 (2017) 149–154
D.A. Nelson et al.
4. Discussion
substantial back pain and depression events that preceded the observed time period, despite our careful examination of total data on the subjects. It is therefore possible that for some subjects, undetected diagnoses of these problems and temporal relationships between them existed. It does not appear possible to estimate the magnitude of this effect, but such information could have biased our results towards the null hypothesis, strengthening our findings. We suggest that the Army's requirements that no serious depression or back pain problems are present among newly-commissioned officers reduce the concern (US Army, 2011). The study was also limited by the inability to assume the findings will extend to the general population or to civilian nurses. We further recognize that the multiple data sources available to the study, such as military-specific sources of BMI data and diagnosis codes for back pain and depression, may not have undergone prior validation in the extant literature. Further research using techniques such as out-of-sample validation would be needed to fully validate our findings. Also, we employed binary indicators for depression and back pain. Specific severity levels of these problems could benefit from further study in order to clarify more subtle relationships that may be present. Additionally, we recognize the potential roles of other factors as back pain predictors, such as anxiety and post-traumatic stress disorder. These factors were not part of this study, but we expect to examine them in ongoing research. Finally, we used retrospective rather than prospective data for this study under an observational approach, methods that can be limited by matters including the absence of a control group. However, the use of these data has provided benefits including the ability to observe natural phenomena in the total target population without bias that might be introduced in experimental designs, such as the Hawthorne effect.
While depression has been associated with back pain (Hoy et al., 2010; Nicholas et al., 2011), the predominant direction of causality has been uncertain. The main finding of this study was a statistically-significant increase in the adjusted odds of later, incident back pain among female RNs when prior depression was present. The extremely-similar effect estimate for the male subpopulation lacked statistical significance, but this relationship deserves further study due to the small male RN group and their attendant low number of outcomes. Another important finding was that the adjusted back pain odds for males and females were approximately doubled when BMI was 30 or greater. The findings help to clarify the impact of BMI on back pain. BMI in the overweight range was associated with back pain in U.S. nurses in one cross-sectional study (Darby et al., 2013), but a casecontrol study of Swedish nurses found no association between back injuries and BMI (Engkvist et al., 2000). Our research supports the former finding. Overall, our findings suggest that careful management of depression and body composition may pay dividends in reducing back pain rates in the nursing profession. The covariates selected for this multivariable analysis were also important in order to adequately address potential confounding. Depression is associated with elevated BMI (Dong et al., 2004), and as noted above, elevated BMI was known to be a potential back pain predictor at the outset of this work. We believe our study is the first to shed light on the potentially-complex temporal relationship between a history of depression and back pain when taking BMI into account, the first to do so using comprehensive, multisource longitudinal data, and the first to do so in the nursing profession. The rates of back pain we discovered in this population were similar to rates reported in literature on the civilian workforce, where U.S. nurses were found to have a 12-month prevalence of 32% (Trinkoff et al., 2003) and a 15-month back pain incidence of 21.1% (Trinkoff et al., 2006). We recognize that these comparisons may be limited by our definition for back pain, which was mostly based on provider diagnoses, whereas the cited studies of civilian nurses used self-report questionnaires. However, the comparisons to civilian rates appear reasonable, as we did include self-report from Periodic Health Assessment data, and overall, there is no definitive, objective test for the presence of back pain. The evaluation of back pain is not well-standardized (Dagenais et al., 2010), and all providers must substantially rely on patient self-report when making their diagnoses. The similarities between prior civilian nurse back pain rates and our findings may be due to relatively few differences between main workplace exposures for civilian and military nurses. While both groups work nominal 40-h work weeks, per communication with the office of the Chief of the US Army Nurse Corps, October 2016, working hours among Army nurses can vary widely along with exposure to manual patient handling training. Similarly, many civilian hospitals have mandatory overtime policies, requiring nurses to stay longer than their usual 12 h shifts. In the absence of specific ergonomic regulations from the Occupational Safety and Health Administration, neither the overall civilian hospital system nor military hospitals (MAMC, 2014) are known to have standardized low lift policies or consistent safe patient handling programs. Combat duty is one unique exposure for the military and its nurses, and it was associated with increases in the adjusted odds of back pain among the RNs we studied. The Army's physical fitness policies and practices (US Army, 2012) that mandate regular exercise and testing for physical performance represent another difference from the civilian workplace, which has no such requirements. However, we find no evidence that the military fitness program itself creates an increased risk for back pain. Whether perioperative work creates increased back pain risk deserves further study, as our sample was sufficiently small to suggest that the finding could have arisen from chance. Other limitations of this work include our inability to rule out
5. Conclusion In conclusion, this study has produced new information on the epidemiology and predictors of back pain among military nurses. Our findings provide actionable occupational health information that might assist military health system leaders in addressing the behavioral health needs of their nurses. The confirmed impact of high BMI on back pain that we found suggests that improved BMI may reduce back pain risk. Studies should be conducted on the impact on back pain rates of military programs such as the Army’s Performance Triad (US Army, 2015b), which has been conceived to help soldiers improve their health, including by reducing BMI. In addition, the Army and civilian health systems might consider standardizing the training of medical personnel in patient handling and the equipment used across facilities in order to reduce overall back pain rates. Such a policy change could provide a positive impact, given that the manual handling of patients is a known back pain risk factor for both military and civilian nurses. Funding This research was supported by an unnumbered seed grant from Leidos, Reston, VA to the University of Maryland. Funders played no role in study design, analysis or the reporting of results. Acknowledgments The authors acknowledge the contributions of the following organizations to our ability to successfully conduct this project: the US Army Office of the Surgeon General, and the Center for Health Informatics and Decision Systems of the Robert H. Smith School of Business, the University of Maryland, College Park. References CDC: Centers for Disease Control and Prevention, 2015. About Adult BMI. http://www.
153
International Journal of Nursing Studies 74 (2017) 149–154
D.A. Nelson et al.
Nelson, D.A., Wolcott, V.L., Kurina, L.M., 2016. Prediction of all-cause occupational disability among US Army soldiers. Occup. Environ. Med. 0, 1–10. http://dx.doi.org/ 10.1136/oemed-2015-103436. Nicholas, M.K., Linton, S.J., Watson, P.J., Main, C.J., 2011. Early identification and management of psychological risk factors (yellow flags) in patients with low back pain: a reappraisal. Phys. Ther. 91 (5), 1–17. Niebuhr, D.W., Krampf, R.L., Mayo, J.A., et al., 2011. Risk factors for disability retirement among healthy adults joining the U.S. Army Mil. Med. 176, 170–175. Patzkowski, J.C., Rivera, J.C., Ficke, J.R., Wenker, J.C., 2012. The changing face of disability in the US army: the operation enduring freedom and operation iraqi freedom effect. J. Am. Acad. Orthop. Surg. 20 (Suppl. 1), 23–30. Pinheiro, M.B., Ferreira, M.L., Refshauge, K., et al., 2015. Symptoms of depression and risk of new episodes of low back pain: a systematic review and meta-analysis. Arthritis Care Res. 67 (11), 1591–1603. Pinheiro, M.B., Ferreira, M.L., Refshauge, K., et al., 2016. Symptoms of depression as a prognostic factor for low back pain: a systematic review. Spine J. 16 (1), 105–116. Rosellini, A.J., et al., 2016. Predicting non-familial major physical violent crime perpetration in the US Army from administrative data. Psychol. Med. 46, 303–316. Shiri, R., Karppinen, J., Leino-Arjas, P., et al., 2010a. The association between obesity and low back pain: a meta-analysis. Am. J. Epidemiol. 171, 135–154. Shiri, R., Karppinen, J., Leino-Arjas, P., Solovieva, S., Viikari-Juntura, E., 2010b. The association between smoking and low back pain: a meta-analysis. Am. J. Med. 123, 87 e7-35. Sikorski, C., Emerson, M.A., Cowan, D.N., et al., 2012. Risk factors for medical disability in U.S. enlisted Marines: fiscal years 2001 to 2009. Mil. Med. 177, 128–134. Trinkoff, A.M., Lipscomb, J.A., Geiger-Brown, J., Storr, C.L., Brady, B.A., 2003. Perceived physical demands and reported musculoskeletal problems in registered nurses. Am. J. Prev. Med. 24 (3), 270–275. Trinkoff, A.M., Le, R., Geiger-Brown, J., Lipscomb, J.A., Lang, G., 2006. Longitudinal relationship of work hours, mandatory overtime, and on-call to musculoskeletal problems in nurses. Am. J. Ind. Med. 49 (11), 964–971. US Army, 2011. Army Regulation 40-501: Standards of Medical Fitness. http://www.apd. army.mil/pdffiles/r40_501.pdf. (Accessed 17 April 2012). US Army, 2012. Field Manual 7-22: Army Physical Readiness Training. http://www.apd. army.mil/pdffiles/fm7_22.pdf. (Accessed 17 April 2012). US Army, 2013. Army Regulation 600-9: The Army Body Composition Program. http:// www.apd.army.mil/pdffiles/ar600_9.pdf. (Accessed 17 April 2012). US Army, 2015a. Medical Operational Data System. https://g1arng.army.pentagon.mil/ Systems/MODS/Pages/default.aspx. (Accessed 17 April 2012). US Army, 2015b. How the Army’s Performance Triad Can Help You Reach Your Healthy Living Goals. http://www.health.mil/News/Articles/2015/03/03/How-the-ArmysPerformance-Triad-Can-Help-You-Reach-Your-Healthy-Living-Goals?type=Articles& page=5#pagingAnchor. (Accessed 16 November 2017). Vermani, E., Mittal, R., Weeks, A., 2010. Pelvic girdle pain and low back pain in pregnancy: a review. Pain Pract. 10 (1), 60–71. Yassi, A., Lockhart, K., 2013. Work-relatedness of low back pain in nursing personnel: a systematic review. Int. J. Occup. Environ. Health 19 (3), 223–244.
cdc.gov/healthyweight/assessing/bmi/adult_bmi/. (Accessed 28 July 2016). Cohen, S.P., Nguyen, C., Kapoor, S.G., et al., 2009. Back pain during war: an analysis of factors affecting outcomes. Arch. Intern. Med. 169, 1916–1923. Cohen, S.P., Brown, C., Kurihara, C., Plunkett, A., Hguyen, C., Strassels, S.A., 2010. Diagnoses and factors associated with medical evacuation and return to duty for service members participating in Operation Iraqi Freedom or Operation Enduring Freedom: a prospective cohort study. Lancet 375, 301–309. Dagenais, S., Tricco, A.C., Halderman, S., 2010. Synthesis of recommendations for the assessment and management of low back pain from recent clinical practice guidelines. Spine J. 10 (6), 514–529. Darby, B., Gallo, A.M., Fields, W., 2013. Physical attributes of endoscopy nurses related to musculoskeletal problems. Gastroenterol. Nurs. 36 (3), 202–208. Davis, K.G., Kotowski, S.E., 2015. Prevalence of musculoskeletal disorders for nurses in hospitals, long-term care facilities, and home health care: a comprehensive review. Hum. Factors 57 (5), 754–792. de Schepper, E.I., et al., 2010. The association between lumbar disc degeneration and low back pain: the influence of age, gender and individual radiographic features. Spine 35 (5), 531–536. Dong, C., Sanchez, L.E., Price, R.A., 2004. Relationship of obesity to depression: a familybased study. Int. J. Obes. 28, 790–795. Engkvist, I.L., Hjelm, E.W., Hagberg, M., Menckel, E., Ekenvall, L., 2000. Risk indicators for reported over-exertion back injuries among female nursing personnel. Epidemiology 11 (5), 519–522. Gaderman, A.M., Engel, C.C., Naifeh, J.A., Nock, M.K., Petkhova, M., Santiago, P.N., Benjamin, W., Zaslavsky, A.M., Kessler, R.C., 2012. Prevalence of DSM-IV major depression among US military personnel: meta-analysis and simulation. Mil. Med. 177 (8), 47–59. Gibbons, R.D., et al., 2003. Waiting for organ transplantation: results of an analysis by an Institute of Medicine Committee. Biostatistics 4 (2), 207–222. Hoge, C.W., Castro, C.A., Messer, S.C., McGurk, D., Cotting, D.I., Koffman, R.L., 2004. Combat duty in Iraq and Afghanistan, mental health problems, and barriers to care. N. Engl. J. Med. 351 (1), 13–22. Hoy, D., Brooks, P., Blyth, F., Buchbinder, R., 2010. The epidemiology of low back pain. Best Pract. Res. Clin. Rheumatol. 24 (6), 769–781. Knox, J., Orchowski, J., Scher, D.L., Owens, B.D., Burks, R., Belmont, P.J., 2011. The incidence of low back pain in active duty United States military service members. Spine 36 (18), 1492–1500. Letvak, S., Ruhm, C.J., McCoy, T., 2012. Depression in hospital-employed nurses. Clin. Nurse Spec. 26 (3), 177–182. Lincoln, A.E., Smith, G.S., Amoroso, P.J., Bell, N.S., 2002. The natural history and risk factors of musculoskeletal conditions resulting in disability among US Army personnel. Work 18, 99–113. MAMC: Madigan Army Medical Center, 2014. MAMC Memorandum Number 385-9. 13 May. Available upon request. Nelson, D.A., Kurina, L.M., 2013. Clinical prediction of musculoskeletal-related ‘Medically Not Ready’ for combat duty statuses among active duty U. S. Army soldiers. Mil. Med. 178, 1365–1372.
154