Seventeen-Year National Pain Prevalence Trends Among U.S. Military Veterans

Abstract

U.S. military veterans experience higher pain prevalence than nonveterans. However, it is unclear how the disparities in pain prevalence have changed over time because previous trend studies are limited to veterans using the Veterans Health Administration. This repeated cross-sectional study aimed to characterize pain prevalence trends in the overall population of U.S. veterans compared to nonveterans, using nationally-representative data. We analyzed 17 years of data from the National Health Interview Survey (2002–2018), with a mean annual unweighted sample of 29,802 U.S. adults (total unweighted n=506,639) and mean annual weighted population of 229.7 million noninstitutionalized adults. The weighted proportion of veterans ranged 11.48% in 2002 (highest) to 8.41% in 2017 (lowest). We found that veterans experience a similar or higher prevalence of pain than nonveterans across the study period, except for severe headache or migraine and facial pain. Pain prevalence among veterans increased over time, with a higher rate of increase compared to nonveterans for all pain variables. From 2002 to 2018 there was an absolute increase (95% CI) in pain prevalence among veterans (severe headache or migraine: 2.0% [1.6% to 2.4%]; facial pain: 1.9% [1.4% to 2.4%]; neck pain: 4.7% [4.1% to 5.2%]; joint pain: 11.4% [10.8% to 11.9%]; low back pain: 10.3% [9.5% to 11.1%]; any pain: 10.0% [9.6% to 10.4%]; and multiple pains: 9.9% [9.2% to 10.6%]. The continued pain prevalence increase among veterans may have implications for healthcare utilization, highlighting the need for improved pain prevention and care programs for this population with a disproportionate pain burden.

INTRODUCTION

The U.S. population included 19.4 million U.S. military veterans in 2020¹ – individuals who served in any military service branch but are no longer serving – approximately 6.9% to 7.7% of the U.S. population since 2011.^2,3 Veterans are disproportionately impacted by multiple chronic health conditions and severe psychological distress compared to age-matched nonveteran U.S. adults^4,5 and more likely to report worse overall health and health-related quality of life.^6–8 They also have higher prevalence of painful conditions than nonveterans, such as doctor-diagnosed arthritis,^6,8,9 and self-reported severe pain.¹⁰

Pain has become more prevalent over the past two decades among U.S. adults overall. A study using Medical Expenditure Panel Survey (MEPS) data from 1997 to 2014 reported care-seeking or disability episodes related to painful conditions increased from 32.9% to 41.1% among U.S. adults.¹¹ Another study using National Health Interview Survey (NHIS) data from 2002 to 2018 reported increasing prevalence of each available pain location across a wide range of sociodemographic subgroups among U.S. adults aged 25–84.¹²

Pain prevalence has also been increasing in veteran-specific populations since the turn of the century. Veterans Health Administration (VHA) data from 2000 to 2007 show a relative 39.7% increase in low back pain (LBP) care.¹³ Additionally, 2004 to 2011 VHA neck and spinal pain prevalence increased from 1.9% to 2.5% and 2.6% to 4.2%, respectively.¹⁴ While these trends are informative, they are limited to VHA-enrolled veterans, who are generally less healthy than veterans receiving care outside the VHA.^15,16 Because of this, these findings are not generalizable to the broader U.S. veteran population; additionally, these studies provide limited ability to compare veteran pain prevalence trends to those reported in the overall U.S. adult population due to differences in the pain variables recorded (and/or their operational definitions), and variability in study period overlap between veteran and nonveteran studies. Without comparable data that are representative of the broader veteran population, it is unclear if the previously identified differences in pain burden between veterans and nonveterans have been static, worsening, or improving over time. Addressing this knowledge gap is essential for insurers (especially veteran-specific insurers like TRICARE), policymakers, and healthcare systems in determining appropriate resource allocation, population-level intervention planning, and/or assessing population-level intervention performance aimed at reducing veteran pain disparities. Therefore, our primary goal was to characterize pain prevalence trends in the overall population of U.S. veterans compared to nonveterans, using nationally representative data. We hypothesized that pain prevalence among veterans has increased over time, but at a higher rate that nonveterans. Furthermore, because of differences in the population burden of pain across demographic groups (e.g., age, sex, race, and ethnicity)^12,17 and the dynamic differences in the demographic make-up between veterans and nonveterans,^18,19 we hypothesized that differences in prevalence trends between veterans and nonveterans would be at least partially related to differences in demographic characteristics.

METHODS

Study Design and Target Population

The NHIS is the largest in-person household health survey in the U.S. and a primary source of key national health indicators.²⁰ This study used data from the NHIS Sample Adult and Person questionnaires from 2002 to 2018.²¹ Additional details regarding sampling design are available from the National Center for Health Statistics (NCHS).²² We present additional NHIS design and target population information relevant to our study in the Supplemental Material (eMethods). The NCHS Research Ethics Review Board reviews and approves NHIS content and methods annually. Interviewers obtain verbal consent for participation from all survey respondents.

We obtained the 2002–2018 NHIS data from IPUMS Health Surveys, which harmonizes the coding of NHIS data variables across survey years to facilitate cross-time comparisons.²³ Data after 2018 were available at the time of analysis. However, the NHIS implemented a major re-design in 2019, which may show differences related to this re-design rather than true changes in trends across survey design periods.²⁴ Some data before 2002 were also available. However, we decided to use data starting in 2002 to limit reporting of trends to the period after September 11, 2001 and the initiation of Operation Enduring Freedom in October 2001 as these were major national events that could impact pain trends differently than ongoing secular trends.

Veteran Status

Our primary exposure of interest was veteran status (veterans vs. nonveterans). Since 2011, the NHIS identifies veterans by asking, “Did you ever serve on active duty in the U.S. Armed Forces, military Reserves, or National Guard?” Before 2011, the NHIS identified veterans by asking, “Have you ever been honorably discharged from active duty in the U.S. Army, Navy, Air Force, Marine Corps, or Coast Guard?;” as a result, National Guard or Reserves veterans nor veterans discharged for misconduct (e.g., general or other than honorable discharge) or courtmartial (e.g., bad conduct or dishonorable discharge) were not included.

Pain Prevalence

The NHIS asks sample adults about pain (yes/no) in four regions that lasted a day or more over the past 3 months (excluding “aches and pains that are fleeting or minor”): severe headache or migraine, facial pain, neck pain, and LBP. Respondents who report experiencing LBP are asked a follow-up question regarding the presence of associated leg pain that “spread down either leg to areas below the knee.” An additional question is asked about joint pain (excluding pain in the joints of the back or neck) in the past 30 days. The structure and wording of these questions (Table 1) is consistent across all NHIS years used in this study.

Table 1.

National Health Interview Survey Sample Adult Pain Variable Ascertainment, 2002–2018

Pain Variable	Question Wording
Severe Headache or Migrainea	“DURING THE PAST THREE MONTHS, did you have… Severe headache or migraine?”
Facial Paina	“DURING THE PAST THREE MONTHS, did you have… Facial ache or pain in the jaw muscles or the joint in front of the ear?”
Joint Painb	“DURING THE PAST 30 DAYS, have you had any symptoms of pain, aching, or stiffness in or around a joint?”
Neck Paina	“DURING THE PAST THREE MONTHS, did you have… Neck pain?”
Low Back Paina	“DURING THE PAST THREE MONTHS, did you have… Low back pain?”
Low Back Pain with Leg Painc	“Did this pain spread down either leg to areas below the knees?”

^aQuestion preceded by the following statement: “The following questions are about pain you may have experienced in the PAST THREE MONTHS. Please refer to pain that LASTED A WHOLE DAY OR MORE. Do not report aches and pains that are fleeting or minor.”

^bQuestion preceded by the following statement: “The next questions refer to your joints. Please do NOT include the back or neck.”

^cAsked as a follow-up question to sample adult respondents reporting LBP during the past 3 months.

Using the five pain variables above, we created two non-mutually exclusive variables. First, we created an “any pain” variable, identifying individuals with ≥1 prevalent pain variable. Second, we created a “multiple pains” variable, identifying individuals with ≥2 prevalent pain variables. While the NHIS provided data regarding pain interference (2016–2017), pain intensity, chronicity of pain, and pain management (2015–2018), we did not include these variables in our analysis because they were not available for our entire study period.²⁵

Statistical Methods

We calculated the crude and adjusted national prevalence (95% confidence intervals [CIs]) of each pain variable among veterans and nonveterans for each quarter from 2002 through 2018^26–28 using SAS 9.4 (Cary, NC). RStudio was used to visualize trends over time by veteran status.^29–31 Analyses incorporated three variables (provided in NHIS data) to account for the survey’s complex, multistage sampling design: primary sampling units, strata, and sample adult weights. The primary sampling weights and strata variables account for sample design clustering and sample design stratification, respectively, to correctly calculate variance when using NHIS data. Sample adult weights account for sample selection probability adjusted for household non-response, age, race, and sex using quarterly Census Bureau population control totals.²¹ Incorporating sample adult weights into statistical analysis is necessary to calculate prevalence point estimates that are representative of the NHIS target population (non-institutionalized US civilian adults). Failing to use statistical software capable of incorporating all three of these variables or failing to incorporate these variables when analyzing NHIS data leads to erroneous results.²¹

Crude prevalence estimates represent the actual observable national-level trends in veterans and non-veterans. Adjusted prevalence estimates standardize demographic characteristics and provide insight into how much of the observed differences in pain prevalence trends between veterans and nonveterans are driven by differences in the demographic compositions of the two groups, including biological differences (e.g., age, sex) and/or differences in lived experience (e.g., racism, or racial and ethnic minoritization). We used respondent-reported age, sex, race, and Hispanic ethnicity as model covariates to account for demographic characteristic differences between veterans and nonveterans. We describe the NHIS coding of demographic characteristic covariates in the Supplemental Material (eMethods). In the context of descriptive epidemiologic studies like ours, adjusted prevalence estimates are helpful for understanding potential reasons for differences between populations (veterans compared to nonveterans) and uncovering hidden population differences driven by variability in covariate distribution.³² However, adjusting away major differences between veterans and nonveterans in our study has the potential to obscure important differences and using adjusted prevalence estimates to make policy and/or resource allocation decisions may lead to reinforcement of actual disparities.^33,34 Because of this, we focus primarily on reporting and interpreting unadjusted results in the main text of our manuscript (as recommended for descriptive epidemiology studies)³⁵ and provide adjusted results in the Supplemental Material.

We used linear regression with a time-by-veteran-status multiplicative interaction term to calculate mean difference in prevalence change over time in each group and associated 95% CIs using robust standard errors. To complement this analysis (which provides insight into differences in the linear trend between veterans and non-veterans across the entire study period), we also provide qualitative descriptions of observed patterns of the plotted trends. We used annualized sample adult weighted prevalence estimates for 2002 and 2018 to calculate the total absolute change (prevalence₂₀₁₈–prevalence₂₀₀₂) and relative change ([prevalence₂₀₁₈–prevalence₂₀₀₂]/prevalence₂₀₀₂) in pain prevalence across the entire 17-year study period for both point estimates and 95% CIs. Based on the extreme rarity of missingness for both veteran status and pain variables, we did not apply any imputation approaches; we excluded individuals missing exposure or outcome data. There were no missing data for age and sex, and missing data in race and Hispanic ethnicity are imputed by NCHS before public data release.³⁶

Although our primary focus was on differences in prevalence change over time, we also calculated the mean differences in pain prevalence between veterans and nonveterans when pooling data across all seventeen years in our study period. This pooled analysis required altering the sample adult weight as recommended to ensure valid estimates.³⁷ Adjusted pooled prevalence difference estimates were adjusted for the same demographic characteristics previously listed.

RESULTS

The mean weighted population between 2002 and 2018 was 229.7 million noninstitutionalized U.S. adults. This ranged from 205.8 million (in 2002) to 249.5 million (in 2018). Veterans accounted for between 8.4% (95% CI: 8.0%, 8.8%; 2017) and 11.5% (95% CI: 11.1%, 11.9%; 2002) of the weighted adult population across the study period (eTable 1, Supplemental Material). Veteran status was missing for a small proportion (≤0.4% [95% CI: 0.2%, 0.5%] annually) of the weighted population due to refusal to answer, failure to ascertain, or respondent (or their proxy) being unsure of veteran status (eTable 2, Supplemental Material). We present a summary of weighted covariates as well as weighted and raw survey sample size by year for veterans and nonveterans in the Supplemental Material (eTable 3 and eTable 4, respectively). Missingness of pain variables was rare (Supplemental Material: eTable 5, eTable 6, eTable 7, eTable 8, eTable 9, eTable 10, eTable 11).

Severe Headache or Migraine

Crude prevalence of severe headache or migraine (Figure 1) ranged from 6.1% (95% CI: 4.1%, 8.1%; 2006, 2^nd quarter) to 12.4% (95% CI: 10.1%, 14.6%; 2011, 2^nd quarter) among veterans and from 12.3% (95% CI: 11.3%, 13.3%; 2007, 2^nd quarter) to 17.7% (95% CI: 16.7%, 18.7%; 2010, 1^st quarter) among nonveterans. Prevalence estimates and trend lines closely approximated when adjusting for demographic differences (eFigure 1, Supplemental Material), with veterans having slightly higher prevalence of severe headache or migraine on average.

Figure 1.

Prevalence Trends by Veteran Status, 2002–2018: Severe Headache or Migraine and Facial Pain.

Annotations: Dotted lines represent initiation of Operation Iraqi Freedom in 2003 and Operation New Dawn in 2010. Yellow shading highlights 2007, when common dips occurred across several pain variables.

Facial Pain

Facial pain (Figure 1) had the lowest prevalence among both veterans and nonveterans. Like severe headache or migraine, crude facial pain prevalence was lower among veterans, ranging from 1.0% (95% CI: 0.3%, 1.7%; 2007, 2^nd quarter) to 5.8% (95% CI: 3.4%, 8.1%; 2018, 1^st quarter) among veterans and from 3.5% (95% CI: 3.0%, 4.1%; 2015, 1^st quarter) to 5.6% (95% CI: 4.9%, 6.3%; 2012, 1^st quarter) among nonveterans. Prevalence estimates and trend lines closely approximated when adjusting for demographic characteristics (eFigure 1, Supplemental Material).

Neck Pain

Prevalence of neck pain (Figure 2) between group appear almost identical up until approximately 2009, where the prevalence trend for veterans appears to diverge and increase at a faster rate through the end of the study period. Crude neck pain prevalence ranged from 10.1% (95% CI: 6.6%, 13.6%; 2008, 4^th quarter) to 20.6% (95% CI: 16.5%, 24.7%; 2017, 4^th quarter) among veterans and from 12.8% (95% CI: 11.3%, 14.2%; 2017, 4^th quarter) to 17.6% (95% CI: 16.1%, 19.1%; 2018, 4^thquarter) among nonveterans. Trend lines showed some separation when adjusting for demographic characteristics (eFigure 2, Supplemental Material), with slightly higher neck pain prevalence among veterans. Adjusted trend lines similarly diverged starting around 2009.

Figure 2.

Prevalence Trends by Veteran Status, 2002–2018: Neck Pain and Low Back Pain.

Annotations: Dotted lines represent initiation of Operation Iraqi Freedom in 2003 and Operation New Dawn in 2010. Yellow shading highlights 2007, when common dips occurred across several pain variables.

Low Back Pain

Prevalence of any LBP (Figure 2) was similar at the start of the study period; however, prevalence quickly began increasing, with trendlines diverging increasingly over time. Crude LBP prevalence ranged from 22.8% (95% CI: 18.2%, 27.4%; 2007, 3^rd quarter) to 40.2% (95% CI: 35.2%, 45.2%; 2015, 4^th quarter) among veterans and from 24.9% (95% CI: 23.7%, 26.6%; 2007, 2^nd quarter) to 30.9% (95% CI: 29.0%, 32.9%; 2018, 4^th quarter) among nonveterans. When adjusting for demographic characteristics (eFigure 2, Supplemental Material), trend lines approximated but showed similar increasing trend divergence over time. We present results from LBP subgroup analyses in the Supplemental Material (eResults, eFigure 3, and eFigure 4).

Joint Pain

Veterans had a notably higher crude joint pain prevalence (Figure 3) across the study period, with no overlap in prevalence estimates with nonveterans in any quarter. This ranged from 34.2% (95% CI: 29.9%, 38.5%; 2007, 1^st quarter) to 52.2% (95% CI: 46.7%, 57.8%; 2018, 4^th quarter) among veterans and from 25.6% (95% CI: 24.1%, 27.1%; 2007, 1^st quarter) to 34.2% (95% CI: 32.2%, 36.2%; 2018, 4^th quarter) among nonveterans. Joint pain prevalence estimates and trend lines closely approximated one another when adjusting for demographic characteristics (eFigure 5, Supplemental Material). These adjusted trends lines showed divergence near the end of the study period, with veterans starting to show slightly higher prevalence estimates.

Figure 3.

Joint Pain Prevalence Trends by Veteran Status, 2002–2018.

Annotations: Dotted lines represent initiation of Operation Iraqi Freedom in 2003 and Operation New Dawn in 2010. Yellow shading highlights 2007, when common dips occurred across several pain variables.

Any Pain

Crude any pain prevalence (Figure 4) ranged from 49.2% (95% CI: 43.6%, 54.9%; 2007, 2^nd quarter) to 64.8% (95% CI: 59.3%, 70.2%; 2018, 4^th quarter) among veterans and from 45.9% (95% CI: 44.2%, 47.6%; 2007, 2^nd quarter) to 54.8% (95% CI: 52.5%, 57.1%; 2009, 1^st quarter) among nonveterans. Differences in any pain prevalence attenuated when adjusting for demographic characteristics (eFigure 6, Supplemental Material), but veterans continued to have higher prevalence than nonveterans across the study period.

Figure 4.

Prevalence Trends by Veteran Status, 2002–2018: Any Pain Complaint and Multiple Pain Complaints.

“Any Pain” is defined by the presence of ≥1 and “Multiple Pains” is defined by the presence of ≥2 of the five primary pain variables available in the NHIS: severe headache or migraine, facial pain, neck pain, joint pain, and low back pain. Annotations: Dotted lines represent initiation of Operation Iraqi Freedom in 2003 and Operation New Dawn in 2010. Yellow shading highlights 2007, when common dips occurred across several pain variables.

Multiple Pains

Crude multiple pain prevalence (Figure 4) was similar between groups at the start of the study period; however, trends diverged around 2003 with higher prevalence of multiple pains among veterans across the second half of the study period. This ranged from 20.8% (95% CI: 16.4%, 25.3%; 2007, 3^rd quarter) to 35.5% (95% CI: 30.7%, 40.2%; 2018, 4^th quarter) among veterans and from 21.6% (95% CI: 19.6%, 23.6%; 2007, 3^rd quarter) to 28.9% (95% CI: 27.1%, 30.7%; 2018, 4^th quarter) among nonveterans. Trend lines showed more separation when adjusting for demographic characteristics (eFigure 6, Supplemental Material), with slightly more pronounced differences between veterans and nonveterans over time. Adjusted prevalence estimates showed similar increasing trend divergence starting around 2003.

Differences in Pain Prevalence Across the Study Period

Our primary focus in this study was on describing trends/changes in pain prevalence across the study period in veterans and compared to nonveterans. We present the absolute and relative change in annual prevalence of each pain variable at the start (2002) and end (2018) of the study period for veterans and nonveterans in the Table 2 and Table 3, respectively. We present the crude and adjusted mean differences in annual prevalence change between veterans and nonveterans in Figure 5.

Table 2.

Change in Annual Pain Prevalence Among U.S. Veterans, National Health Interview Survey 2002–2018

	Annual Prevalence		Total Annual Prevalence Change, 2002 to 2018a
Pain Variable	2002	2018	Absolute Change	Relative Change
Severe Headache or Migraine	8.3% (95% CI: 7.4%, 9.3%)	10.3% (95% CI: 9.0%, 11.7%)	2.0% (95% CI: 1.6%, 2.4%)	24.2% (95% CI: 21.8%, 26.0%)
Facial Pain	2.7% (95% CI: 2.2%, 3.3%)	4.6% (95% CI: 3.6%, 5.7%)	1.9% (95% CI: 1.4%, 2.4%)	69.4% (95% CI: 66.4%, 71.3%)
Neck Pain	14.1% (95% CI: 12.8%, 15.4%)	18.7% (95% CI: 16.9%, 20.6%)	4.7% (95% CI: 4.1%, 5.2%)	33.2% (95% CI: 32.2%, 34.0%)
Joint Pain	38.3% (95% CI: 36.5%, 40.0%)	49.6% (95% CI: 47.3%, 51.9%)	11.4% (95% CI: 10.8%, 11.9%)	29.7% (95% CI: 29.6%, 29.8%)
Low Back Pain	26.5% (95% CI: 24.9%, 28.1%)	36.8% (95% CI: 34.4%, 39.2%)	10.3% (95% CI: 9.5%, 11.1%)	38.7% (95% CI: 38.0%, 39.4%)
Associated Leg Pain	8.6% (95% CI: 7.6%, 9.7%)	15.6% (95% CI: 13.8%, 17.4%)	6.9% (95% CI: 6.3%, 7.6%)	80.4% (95% CI: 82.9%, 78.5%)
No Associated Leg Pain	17.8% (95% CI: 16.4%, 19.2%)	21.1% (95% CI: 19.2%, 23.1%)	3.3% (95% CI: 2.7%, 3.9%)	18.7% (95% CI: 16.8%, 20.3%)
Any Pain	53.6% (95% CI: 51.8%, 55.4%)	63.6% (95% CI: 61.4%, 65.8%)	10.0% (95% CI: 9.6%, 10.4%)	18.7% (95% CI: 18.5%, 18.9%)
Multiple Pains	23.7% (95% CI: 22.1%, 25.3%)	33.6% (95% CI: 31.2%, 35.9%)	9.9% (95% CI: 9.2%, 10.6%)	41.7% (95% CI: 41.6%, 41.8%)

Annual prevalence estimates calculated using NHIS annualized sample adult weights from NHIS 2002 and 2018. Absolute change from 2002 to 2018 calculated using annual prevalence estimates. Denominator for each pain variable includes all weighted sample adults who were missing neither veteran status nor the relevant pain variable due to refusal to answer, failure to ascertain, or respondent (or their proxy) being unsure of veteran or pain variable status. “Any Pain” is defined by the presence of ≥1 and “Multiple Pains” is defined by the presence of ≥2 of the five primary pain variables available in the NHIS: severe headache or migraine, facial pain, neck pain, joint pain, and low back pain.

^aApparent mismatch in change calculations are related to rounding.

Table 3.

Change in Annual Pain Prevalence Among U.S. Nonveterans, National Health Interview Survey 2002–2018

	Annual Prevalence		Total Annual Prevalence Change, 2002 to 2018a
Pain Variable	2002	2018	Absolute Change	Relative Change
Severe Headache or Migraine	16.0% (95% CI: 15.5%, 16.6%)	16.0% (95% CI: 15.3%, 16.6%)	0.0% (95% CI: −0.1%, 0.1%)	−0.1% (95% CI: −0.7%, 0.4%)
Facial Pain	4.9% (95% CI: 4.6%, 5.2%)	5.2% (95% CI: 4.8%, 5.6%)	0.3% (95% CI: 0.3%, 0.4%)	6.8% (95% CI: 6.5%, 7.1%)
Neck Pain	13.8% (95% CI: 13.3%, 14.4%)	15.8% (95% CI: 15.2%, 16.5%)	2.0% (95% CI: 1.9%, 2.2%)	14.7% (95% CI: 14.2%, 15.2%)
Joint Pain	28.4% (95% CI: 27.8%, 29.1%)	32.8% (95% CI: 31.9%, 33.8%)	4.4% (95% CI: 4.2%, 4.7%)	15.5% (95% CI: 15.0%, 16.0%)
Low Back Pain	26.5% (95% CI: 25.8%, 27.1%)	29.2% (95% CI: 28.3%, 30.1%)	2.7% (95% CI: 2.5%, 3.0%)	10.3% (95% CI: 9.7%, 10.9%)
Associated Leg Pain	8.2% (95% CI: 7.8%, 8.6%)	10.4% (95% CI: 9.8%, 10.9%)	2.2% (95% CI: 2.0%, 2.3%)	26.4% (95% CI: 26.1%, 26.7%)
No Associated Leg Pain	18.2% (95% CI: 17.7%, 18.8%)	18.8% (95% CI: 18.1%, 19.5%)	0.6% (95% CI: 0.4%, 0.8%)	3.2% (95% CI: 2.3%, 4.1%)
Any Pain	49.0% (95% CI: 48.2%, 49.7%)	52.9% (95% CI: 51.9%, 53.9%)	4.0% (95% CI: 3.7%, 4.2%)	8.1% (95% CI: 7.7%, 8.4%)
Multiple Pains	24.3% (95% CI: 23.6%, 25.0%)	27.3% (95% CI: 26.4%, 28.1%)	3.0% (95% CI: 2.8%, 3.2%)	12.4% (95% CI: 12.0%, 12.7%)

Annual prevalence estimates calculated using NHIS annualized sample adult weights from NHIS 2002 and 2018. Absolute change from 2002 to 2018 calculated using calculated annual prevalence estimates. Denominator for each pain variable includes all weighted sample adults who were missing neither veteran status nor the relevant pain variable due to refusal to answer, failure to ascertain, or respondent (or their proxy) being unsure of veteran or pain variable status. “Any Pain” is defined by the presence of ≥1 and “Multiple Pains” is defined by the presence of ≥2 of the five primary pain variables available in the NHIS: severe headache or migraine, facial pain, neck pain, joint pain, and low back pain.

^aApparent mismatch in change calculations are related to rounding.

Figure 5.

Mean Difference in Annual Change in Pain Prevalence Between Veterans and Nonveterans, 2002–2018.

“Any Pain” is defined by the presence of ≥1 and “Multiple Pains” is defined by the presence of ≥2 of the five primary pain variables available in the NHIS: severe headache or migraine, facial pain, neck pain, joint pain, and low back pain. Adjusted differences are adjusted for respondent-reported age, sex, race, and ethnicity.

We present the crude and adjusted 17-year pooled differences in pain prevalence between veterans and nonveterans (Prevalence_veterans – Prevalence_nonveterans) in the Supplemental Material (eResults). These pooled differences varied in size, with some pooled differences being less than 1%. However, because these estimates pool NHIS data across almost two decades, they hide nuances presented above in our results (such as the changes within these groups across the study period and how those changes compare over time) and do not accurately represent contemporary group differences in prevalence (i.e., difference in pain prevalence at the end of the study period).

DISCUSSION

We investigated pain prevalence trends among U.S. military veterans compared to nonveterans from 2002 to 2018. The prevalence of most pain variables across the study period was either similar or higher among veterans compared to nonveterans. Further, all pain variables had larger prevalence increases across the 17-year study period among veterans, even after adjusting for demographic characteristics. These findings suggest that pain prevalence increased at a higher rate on average among veterans over this period and the different rate increase is not simply attributable to demographic differences compared to nonveterans.

Annual multiple pain prevalence grew 9.9% (41.7% relative increase) among veterans; an increase that was 3.3-times higher than nonveterans. Similarly, prevalence of having ≥1 of the 5 pain variables was higher among veterans, with annual prevalence increasing 10.0% (18.7% relative increase). This was 2.5-times higher than the increase among nonveterans. Similarly, another study using 18 years of MEPS data reported an 8.1% absolute (25.0% relative) increase in the prevalence of any noncancerous painful health condition in the past year resulting in a medical event or disability episode.¹¹ Our study builds onto this evidence by comparing veterans and nonveterans and capturing pain not limited by its connection to care-seeking or disability episodes.

Joint pain in veterans had the largest absolute increase in annual prevalence (11.4%); a 2.6-times higher total increase than nonveterans. A study of self-reported doctor-diagnosed osteoarthritis using nationally-representative data from the National Health and Nutrition Examination Survey (NHANES) years 2005–2018 also reported increasing age-adjusted prevalence trends.³⁸ But, estimates were not presented by veteran status and reported prevalence was lower than our estimates of joint pain. The NHIS definition of joint pain in our study is not specific to doctor-diagnosed arthritis like in NHANES.

Neck and LBP trends in a study using all VHA data from 2004 to 2011 showed an increase from 1.9% to 2.5% (31.6% relative increase), 12.3% to 16.2% (31.7% relative increase), and 2.6% to 4.2% (61.5% relative increase) in veterans seeking care for neck, back, and spine pain in multiple spine segments, respectively.¹⁴ Prevalence of neck and LBP in our study were higher and had larger growth than trends from VHA data. These features are likely related to our study’s longer study period and more inclusive pain experiences that did not require a care-seeking encounter. Capturing pain regardless of VHA use in our study is a major strength, as only 30.4% of U.S. veterans used the VA for healthcare in 2019.¹⁶

Severe headache or migraine and facial pain had the lowest prevalence among veterans and – differently than other pain variables – were less prevalent than in nonveterans across the study period. Despite lower prevalence among veterans, our results showed larger increases in prevalence for severe headache or migraine (2.0% absolute increase; 24.2% relative increase) and facial pain (1.9% absolute increase; 69.4% relative increase) among veterans. In comparison, nonveterans had marginal changes over this same period. Interestingly, an increasing trend in migraine and non-migraine headaches has also been reported among active duty U.S. military personnel using military healthcare system data from 1998–2010, potentially coinciding with more military deployments from 2005–2010.³⁹ When we adjusted for quarterly demographic characteristics, prevalence was slightly higher among veterans for both facial pain and severe headache or migraine. So, while fewer U.S. veterans have severe headache or migraine and facial pain, they appear to be slightly more burdened by these conditions than nonveterans with similar demographic characteristics. This reversal between crude and adjusted estimates is likely due to the greater proportion of males and higher mean age in the U.S. veteran population, given the strong female predominance (approximately 2:1) of severe headache or migraine and facial pain, and lower prevalence of these conditions after middle age.³⁹

We observed notable dips in prevalence that occurred in 2007 among veterans for several pain variables. Nonveterans had a similar but less pronounced dip for some pain variables during the same year and demonstrated a similar 2007 dip in prevalence for severe headache and migraine not observed among veterans. The dips may be related to an NHIS sampling redesign implemented in 2006²⁷ and budget shortfalls that reduced the targeted annual sample size from 2006–2008.⁴⁰ The dips among veterans may be further related to a decrease of 458 thousand veterans in the total weighted population between 2006 and 2007. This reduction could be a result of voluntary (e.g., high reenlistment bonuses peaking in fiscal years 2006 and 2007) and involuntary military retention policies (i.e., Stop Loss) in place at that time.^41,42 This timing also coincides with tour extension for many units (from 12 to 15 months) and escalation of troops overseas starting in 2007.^42–44 However, the exact reasons for these dips are unclear.

Adjusting prevalence for quarterly demographic characteristics attenuated veteran and nonveteran trendlines to varying degrees for severe headache or migraine, facial pain, joint pain, and LBP. This attenuation indicates that at least some of the within-quarter prevalence differences for these variables were related to demographic characteristic differences between veterans and nonveterans. On the other hand, demographic adjustment for neck pain and multiple pain prevalence resulted in a slight separation of trendlines, indicating differences in the prevalence of these two types of pain would be even bigger than those observed if the demographic make-up of veterans were the same as nonveterans. While these adjusted estimates provide insights, crude estimates in our study represent actual veteran and nonveteran demographic composition, and thus have more relevance for policymakers and are more helpful when considering resource allocation.^32,35

Even though some trendlines came closer to one another on the Y-axis after adjusting for demographic characteristics and others separated, the overall shape of the trendlines remained similar. Our results showed minimal to no impact of demographic adjustment on the mean differences in annual prevalence change from 2002 to 2018 – higher growth in annual pain prevalence was largely unrelated to demographic characteristic differences between veterans and nonveterans. An exhaustive analysis of other NHIS variables that may help explain the higher rates is beyond the scope of our current work, but because veterans on average have higher educational attainment, median individual and household income, proportion of home and vehicle ownership, and are more likely to have health insurance coverage,^45–49 it seems unlikely that other indicators of socioeconomic advantage/disadvantage would account for the differences in pain prevalence growth. Changes in military service era make-up and associated experiences over time among veterans may be related, as post-9/11 veterans experienced more deployments, more combat, more emotional trauma, post-traumatic stress, and more difficulty readjusting to civilian life than their pre-9/11 counterparts.⁵⁰ Data were not available to explore this potential relationship between service-related characteristics or service-connected conditions (physical injuries or otherwise) and the growth rate of prevalence among veterans.

We are unaware of direct comparisons of healthcare utilization trends for pain between veterans (regardless of their VHA use) and nonveterans. Prior studies indicate that pain is one of the most expensive health conditions in the U.S.^51,52 Thus, we expect that if the disparate upward trends in pain prevalence observed from 2002 to 2018 were to continue, they would likely come with similar disparate trends in associated healthcare needs. This concern is compounded by evidence that veterans experience higher pain severity than nonveterans,¹⁰ which is associated with higher healthcare utilization.⁵³

Limitations

Our study provides estimates representative of the overall non-institutionalized U.S. veteran population, receiving care within and outside of the VHA. While this is a strength, VHA enrollment status of veteran respondents is unknown for our study period because the NHIS did not collect this variable before 2019. Future studies using data from the NHIS 2019 redesign and beyond will be able to investigate these veteran subgroups further. Likewise, the NHIS does not collect additional military service characteristics from veterans that may be of interest (e.g., combat exposure, number and duration of deployments). The NHIS excludes older individuals who are living in long-term care institutions, adults of any age living in correctional facilities, and U.S. nationals living abroad; limiting generalizability to those veterans. Pain questions used in our study asked respondents about pain during the past 3 months (or 30 days for joint pain). The wording limits determination of chronicity, intensity, or interference or persistence. Further, the NHIS is an annually repeated cross-sectional survey; although the target population of interest is the same across all years, the NHIS does not follow respondents longitudinally, preventing individual-level inferences from year-to-year.

Conclusions

This study is the first to report trends in pain prevalence among veterans outside of the VHA that are directly comparable to nonveterans in terms of representativeness and pain variable ascertainment. Both absolute and relative total increases for all pains were larger among veterans than nonveterans over the 17-year study period. Veterans had higher rates of increase over time and had similar or higher prevalence of all pain variables compared to nonveterans with similar demographic characteristics. Continued accelerated increase in pain among veterans would likely impact healthcare utilization (within and outside of the VHA) and underscores the need for improved pain prevention and care programs for these individuals with disproportionate pain burden.

Perspective:

This article uses routinely-collected cross-sectional data that are nationally-representative of U.S. adults to present changes in pain prevalence among military veterans compared to nonveterans. The findings underscore the need for improved prevention and pain care programs for veterans, who experienced a widening disproportionate pain burden from 2002 to 2018.

HIGHLIGHTS.

• Pain disproportionately burdens U.S. military veterans

• Veterans had higher rates of increasing pain prevalence from 2002 to 2018

• Disparities persist when accounting for key demographic characteristic differences