Combat Experience, New-Onset Mental Health Conditions, and Posttraumatic Growth in U.S. Service Members

Abstract

Objective: Studies examining posttraumatic growth (PTG) rely on surveys evaluating PTG in relation to prior traumatic experiences, resulting in psychometric problems due to the linkage of the dependent and independent variables. Few studies have assessed PTG following combat deployment while also controlling for mental health problems.

Method: Longitudinal data on PTG, combat experience, and mental health were examined among U.S. Millennium Cohort Study deployers (n = 8732). Scores from a short-form (SF) version of the PTG inventory assessing current-state beliefs (C-PTGI-SF) independent of any predictor variables were assessed at time 1 (T1), before deployment, and change in scores were assessed approximately 3 years later after deployment at time 2 (T2). All participants screened negative for posttraumatic stress disorder (PTSD) and depression at T1.

Results: Combat deployment severity was associated with a worsening of C-PTGI-SF scores at T2 among participants with moderate C-PTGI-SF scores at T1. A positive screen for comorbid PTSD/depression was associated with a worsening of C-PTGI-SF scores at T2 among participants with moderate or high C-PTGI-SF scores at T1. At T2, a strong inverse correlation was found between C-PTGI-SF scores and PTSD (r = −0.38) and depression (−0.41). Only 5% of participants who screened positive for a mental health problem at T2 (23/517) also experienced positive growth.

Conclusions: These results challenge the clinical utility of the PTG construct. While PTG may be a useful framework for supporting trauma recovery on an individual basis, PTG does not appear to be distinct and independent from the negative psychological impact of traumatic experiences.

The negative psychological effects of combat exposure among U.S. service members deploying to operations in Iraq and Afghanistan have been well documented over the last two decades (Hoge, 2006; Milliken et al., 2007; Smith et al., 2008), while reports on the association of posttraumatic growth following combat experience are relatively sparse. Tedeschi and Calhoun (1996) developed the term posttraumatic growth (PTG) to describe a construct positing that the trauma of a tragic event may be associated with resilience and growth (positive psychological improvement), and this growth may enable better psychological adjustment (Tedeschi & Calhoun, 1996). A strength of the PTG construct is that it acknowledges that the processes involved in rebuilding one’s reality or new sense of normalcy after a trauma can lead to growth (Tedeschi & Calhoun, 2004), and this growth may be experienced across multiple domains (e.g., “relating to others,” “new possibilities”, “personal strength”, “spirituality”, or “appreciation of life”). It may be, however, that rather than describing genuine growth, the PTG construct reflects the consequences of a self-protective mechanism. That is, rather than focus on the detrimental impact of the negative experience, individuals focus on benefits they perceive as resulting from a traumatic event, and this focus serves what Zoellner and Maercker (2006) refer to as “self-illusory coping”. Little research has examined whether PTG interrelates with mental health problems. Research into the PTG construct is needed to clarify these dynamics given their implications for early intervention and treatment.

To date, most research on PTG has relied on the Posttraumatic Growth Inventory (PTGI) (Tedeschi & Calhoun, 1996), a scale designed to assess the construct by retrospectively measuring the extent to which people experience changes after a trauma. A recent review article highlighted several publications that evaluated positive growth in relation to military service (Mark et al., 2018), most of which used the PTGI. In terms of combat, four of the studies reviewed by Mark et al. (2018) found a positive association between combat experience and PTG (Bush et al., 2011; Gallaway et al., 2011; Mitchell et al., 2013; Park et al., 2017), one study found a negative association (McLean et al., 2013), and five studies found non-significant associations (Currier et al., 2013; Kaler et al., 2011; Marotta-Walters et al., 2015; Pietrzak, 2010; Tsai et al., 2015). Of these 10 studies, only one had a longitudinal design (Kaler et al., 2011) but was assigned a “poor” quality rating, emphasizing the need for longitudinal work in this area. In addition, only 2 studies have examined combat experience as a predictor of PTG.

Although the PTG theory suggests that growth may occur following a traumatic event, there are significant measurement issues. The original PTGI asks respondents to “Indicate for each of the statements below the degree to which this change occurred in your life as a result of your crisis, using the following scale” with response options ranging from “I did not experience this change as a result of my crisis” to “I experienced this change to a very great degree as a result of my crisis.” Unfortunately, responses may be impacted by retrospective bias when describing perceived changes, and the likelihood of attribution bias is high since independent and dependent variables are conflated in each question that asks individuals to attribute life changes to a specific “crisis” event.

An additional issue with PTG is the lack of systematic research determining the degree to which the construct of PTG is associated with or independent from the negative impacts of trauma, such as posttraumatic stress disorder (PTSD) and depression. Tedeschi and Calhoun (2004) posit that growth following trauma is “far more common than psychiatric disorders”, yet psychiatric distress and growth may co-occur. The relationship between PTG and mental distress is not clear cut and likely depends on the severity of the trauma and how acute reactions transition to long-term coping or recovery (Tedeschi & Calhoun, 2004). Previous research examining PTG and either PTSD or depression has produced mixed results. Studies reviewed by Mark et al. (2018) found positive (Morgan & Desmarais, 2017; Morgan et al., 2017; Pietrzak, 2010; Tsai et al., 2015; Tsai & Pietrzak, 2017), partially positive (Marotta-Walters et al., 2015), negative (Bush et al., 2011; Park et al., 2017; Tsai et al., 2016), and no association of PTSD with PTG (Currier et al., 2013; Gallaway et al., 2011; Kaler et al., 2011; Murphy, 2017). Similarly inconsistent results were observed for depression; PTG was negatively associated with depression in two studies (Bush et al., 2011; Ga et al., 2016) and not significantly associated in four studies (Currier et al., 2013; Gallaway et al., 2011; Kaler et al., 2011; Murphy, 2017). Other studies have examined PTG as a predictor of subsequent mental health (Tsai et al., 2016). In a separate study, Tsai and Pietrzak (2017) identified three trajectories for PTG and showed that greater PTSD severity was associated with membership in the “moderate” or “high and increasing” PTG trajectory classes. While helpful in revealing distinct trajectories of PTG, this study was limited by the fact that 1) only lifetime traumatic events occurring any time prior to the T1 assessment were included although such events could have occurred during follow-up, and 2) those entering the study with greater PTSD symptom severity may have been more likely to endorse a positive outlook to facilitate coping.

Given the psychometric measurement problems in both the full and short-form PTGI (PTGI-SF), some studies have sought to assess the scientific validity of the PTGI, and whether participants completing these items experienced actual change, or whether their perceived changes were due to recall or attributional bias (Frazier et al., 2009; Nolen-Hoeksema & Davis, 2004; Park & Sinnott, 2018). To address these issues, Frazier et al. (2009) adapted the PTGI to a “current standing” instrument (C-PTGI), by asking participants about their feelings over the past two weeks and repeating these questions eight weeks later. Traumatic events were measured using the Traumatic Life Events Questionnaire (Kubany, 2004) and included items such as “sudden and unexpected death of a close friend or loved one” and “any other event that was life-threatening, caused serious injury, or was highly distressing.” “Perceived growth” was measured using the PTGI while “actual growth” was measured using the C-PTGI. In their longitudinal study, Frazier and colleagues found no appreciable relationship between perceived growth (PTGI scores) and actual growth (change in C-PTGI); additionally perceived growth was associated with an increase in depression/anxiety while actual growth was associated with a decrease in depression/anxiety. This study was limited to a sample of undergraduate students where less than one-third (27%) of the eligible sample experienced a traumatic event during the eight-week follow-up period, a time frame which may have been insufficient for growth to occur. Furthermore, there was no control for mental health distress as a potential covariate accounting for any associations between trauma exposure and PTG. Nevertheless, this work serves as a foundational methodological improvement underscoring the importance of a longitudinal approach and adapting the PTGI to assess current state beliefs, minimizing potential retrospective recall or attributional bias. Frazier’s work was further advanced by Johnson and Boals (2015) who replicated Frazier’s methodology, but found that trauma responses varied by “event centrality” (i.e. “the extent to which an individual interprets a stressful event as part of their core identity”). Specifically, trauma with low event centrality produced small correlations between PTGI scores and “actual growth” (measured using the C-PTGI and six functional health constructs), while trauma with high event centrality produced strong correlations between PTGI scores and “actual growth”, suggesting that traumatic events that fundamentally change a person’s sense of self or normalcy may enable higher levels of growth.

To assess PTG over time, the present study analyzed data from the Millennium Cohort Study, a large Department of Defense study of service members and Veterans (Ryan et al., 2007; Smith, 2009). A short version of the C-PTGI (C-PTGI-SF) (Cann et al., 2010; Frazier et al., 2009; Kaur et al., 2017) was added to both the 2013 and 2016 data collections, enabling examination of well-being at each time point and its relationship to PTSD and depression. Building on Frazier et al.’s methodology (2009), this study expanded the focus to individuals at risk for exposure to potentially traumatic events: service members deployed in support of military operations in Iraq and Afghanistan.

The goal of the present study was to examine the relationships between PTG, trauma exposure, and mental health. Specifically, the research questions were: 1) Among deployers, did combat severity, above and beyond mental health problems, lead to PTG approximately 3 years later, compared with service members who deployed and did not experience combat? 2) Was there a correlation between PTSD and PTG, and did those who developed PTSD also experience PTG, controlling for combat severity? and 3) Was there a correlation between depression and PTG, and did those who developed depression also experience PTG, controlling for combat severity?

MATERIALS AND METHOD

Participants

Data for this analysis were drawn from the Millennium Cohort Study (Ryan et al., 2007), which currently has 201,619 participants enrolled into four study panels between 2001 and 2013. Participants represent all service branches, including the Reserves and National Guard, and are surveyed every three to five years. Deployment status was determined using electronic military records from the Defense Manpower Data Center (DMDC). For this study, eligible participants completed two consecutive surveys via web or paper at Time 1 ([T1]: 2011–2013) and Time 2 ([T2]: 2014–2016), completed a deployment between T1 and T2, were still actively serving at T2, and screened negative for PTSD and depression at T1 (n = 10,010). Among eligible participants, those who did not complete any PTG items (n = 961), were missing primary combat-exposure items (n = 147), or had missing covariate data (n = 170) were excluded. The resulting study population included 8,732 individuals. Participants voluntarily provided informed consent at study enrollment. The study was approved by the Naval Health Research Center institutional review board (protocol number NHRC.2000.0007).

Procedure

Measures

The C-PTGI-SF (Kaur et al., 2017) administered at T1 and T2 included a total of 11 items (Cronbach’s alpha was .90 at T1 and .91 at T2): the original 10 PTGI-SF items and one additional item about compassion for others (see Table 1 for items). The question stem was modified by removing the reference to a specific trauma, and asked participants to “Indicate the degree to which the following statements are true in your life … .” Consistent with the original measure, response options were on a 6-point Likert scale (0 = “Not at all” to 5 = “To a very great degree”). Scores were summed such that higher scores indicated more positive perspective (range, 0–55). The C-PTGI-SF scores from T1 were grouped into categories labeled “low,” (range, 0–33.5) “moderate,” (range, 33.6–50.6) and “high” (50.7–55.0), based on their relationship of being below, within, or above one standard deviation from the T1 sample mean (mean = 42.1, SD = 8.49). For modeling purposes, change scores were created and categorized by subtracting T1 scores from T2 scores. Participants were considered to have made either a “positive change” (increasing more than one T1 standard deviation, > 8.49), “negative change” (decreasing more than one T1 standard deviation, < −8.49), or “no change” relative to T1 (−8.49 ≤ change score ≤ 8.49).

TABLE 1. Items from the C-PTGI-SF Instrument on the Millennium Cohort Study Survey, 2013 and 2016

Question stem: Indicate the degree to which the following statements are true in your life …

Response options: 0) Not at all, 1) To a very small degree, 2) To a small degree, 3) To a moderate degree, 4) To a great degree, 5) To a very great degree

a. I prioritize what is important in life

b. I have an appreication for the value of my own life

c. I am able to do good things with my life

d. I have an understanding of spiritual matters

e. I have a sense of closeness with others

f. I have established a path for my life

g. I know that I can handle difficulties

h. I have religious faith

i. I’m stronger than I thought I was

j. I have learned a great deal about how wonderful people are

k. I have compassion for others

The primary exposure to trauma was based on 13 combat events during deployment in support of the operations in Iraq or Afghanistan (e.g., being attacked or ambushed, receiving small arms fire). The items were adapted from the Land Combat Study (Hoge et al., 2004; Kim et al., 2012) (Cronbach’s alpha was .82 in this sample). Items were assessed at T2 as having occurred during the same 3-year period as their deployment between T1 and T2. Response options included “no”, “1 time” or “more than 1 time”. Each exposure was coded as either present (responses of “1 time” or “more than 1 time”) or absent (response of “no”). Combat deployment severity was based on the number of items reported: 0 = deployed without combat, 1–3 = deployed with low combat severity, and ≥ 4 = deployed with medium/high combat severity (Armenta et al., 2018).

The validated 17-item PTSD Checklist−Civilian Version (PCL-C) (Blanchard et al., 1996) was used to screen for PTSD (Cronbach’s alpha was .89 at T1 and .95 at T2). Participants who screened positive for newly-reported PTSD endorsed a moderate or greater level (on a Likert scale scored 1 to 5) of at least one intrusion symptom, three avoidance symptoms, and two hyperarousal symptoms at T2 based on criteria from the Diagnostic and Statistical manual of mental disorders, 4^th edition (DSM-IV-TR; American Psychiatic Association, 2000).

The validated 8-item Patient Health Questionnaire (PHQ-8) was used to screen for depression (Spitzer et al., 1999, 1994) (Cronbach’s alpha was .80 at T1 and .89 at T2). Based on the DSM-IV major depressive disorder criteria, participants screened positive for newly-reported major depressive disorder if they endorsed at least five of the PHQ items as “more than half the days” or “nearly every day” on a 4 point Likert scale, including the item regarding anhedonia or depressed mood. Participants screening positive for both disorders were classified as having newly-reported comorbid PTSD and depression.

Covariates

Covariates were selected a priori based on their potential to influence the relationship between PTG and combat deployment severity or mental health problems. Data obtained from DMDC included date of birth (to derive age group), sex, race/ethnicity, service branch, service component, occupation, and paygrade (see Supplemental Table1). Covariate data from the T1 survey included: marital status, education level, smoking status, problem drinking (defined as responding “yes” to any 1 of 5 items from the PHQ module on alcohol use; Spitzer et al., 1999, 1994), and social support (bothered versus not bothered), assessed as being bothered by “having no one to turn to” based on a PHQ item (Spitzer et al., 1999, 1994). Any previous deployment(s) with combat experiences prior to T1 was assessed using a combination of survey and DMDC deployment data.

Data Analysis

Descriptive analyses included 1) comparison of distributions of C-PTGI-SF scores at T1 (i.e. low, moderate, high) by combat deployment severity and mental health problems (newly-reported PTSD and/or depression) stratified by change in C-PTGI-SF scores at T2 (i.e. positive change, negative change, or no change) and 2) examination of Pearson correlation coefficients between combat deployment severity and mental health problems with C-PTGI-SF scores (Gignac & Szodorai, 2016).

Prior to model building, variance inflation factors were examined with a threshold factor ≥ 4 to indicate collinearity (Harrell et al., 1996). Separate multivariable logistic regression models were fit to determine which variables were significantly associated with categories of change (p < .05). Analyses were stratified by T1 C-PTGI-SF categories since cut points at T1 precluded certain directions of change. Binary logistic regression was used in the T1 “low” group to estimate the odds ratio (OR) of a positive change versus no change (or negative change) and in the “high” group at T1 to estimate the OR of negative change versus no change (or positive change). For those in the T1 “moderate” group, polychotomous logistic regression was employed to assess positive or negative change versus no change.

To determine whether combat deployment severity was associated with change in C-PTGI-SF scores independent of mental health problems, multivariable models were adjusted for the presence or absence of newly-reported mental health problems at T2 and covariates (See Supplemental Table 2 for covariate results).

Results from models using complete-case analyses were compared with results from models using multiple imputation procedures that imputed missing data for C-PTGI-SF items in the adjusted models. Complete-case model results are shown, as there was no appreciable difference between results from analyses of imputed data. Data analyses were conducted using SAS, version 9.4 (SAS Institute, Inc., Cary, North Carolina).

RESULTS

The sample was predominantly male (79%), under 35 (71%), White non-Hispanic (78%), less than college educated (55%), married (68%), Air Force (47%) or Army (37%), enlisted rank (63%), and active duty (76%), and 58% reported combat experiences between T1 and T2. Participants were relatively healthy at T1, and reported no problematic drinking (93%), never smoking (67%), and comfort with their level of social support (88%); 6% screened positive for a newly-reported mental health problem at T2 (1% for depression, 3% for PTSD, and 2% for comorbid conditions). Participant characteristics by PTG grouping at T1 are listed in Supplemental Table 1.

First, we examined unadjusted changes in C-PTGI-SF and combat deployment severity between T1 and T2. Among participants with low C-PTGI-SF scores at T1, a higher proportion of those with no combat experiences between T1 and T2 had a positive change (31.7%) compared with those reporting low combat severity (27.1%) or medium/high combat severity (23.4%) (see upper section of Table 2). Among those with either moderate or high C-PTGI-SF scores at T1, a greater proportion of individuals with low or medium/high combat severity had a negative change in C-PTGI-SF score compared with those with no combat experiences.

TABLE 2. Combat Deployment and Mental Health Status by C-PTGI-SF Scores at T1, Stratified by Change in C-PTGI-SF Scores at T2

	Low PTG at T1 (n = 1,320)				Moderate PTG at T1 (n = 5,902)						High PTG at T1 (n = 1,510)
	No change or negative change (n = 951)		Positive change (n = 369)		Negative change (n = 663)		No change (n = 4,713)		Positive change (n = 526)		Negative change (n = 348)		No change (n = 1162)
Characteristic	N	Row %	N	Row %	N	Row %	N	Row %	N	Row %	N	Row %	N	Row %
Combat Severity
Deployed, no combat	357	68.3	166	31.7	215	8.9	1988	81.8	227	9.3	153	22.7	521	77.3
Deployed, low combat	325	72.9	121	27.1	236	12.0	1547	79.0	176	9.0	106	22.8	358	77.2
Deployed, medium/high combat	269	76.6	82	23.4	212	14.0	1178	77.9	123	8.1	89	23.9	283	76.1
Mental Health Problems
None	809	69.4	357	30.6	518	9.2	4573	81.6	515	9.2	307	21.3	1136	78.7
Depression only	32	91.4	3	8.6	21	38.9	28	51.9	5	9.3	7	63.6	4	36.4
PTSD only	59	90.8	6	9.2	72	47.4	75	49.3	5	3.3	19	52.8	17	47.2
Comorbid PTSD and depression	51	94.4	3	5.6	52	57.8	37	41.1	1	1.1	15	75.0	5	25.0

Second, we examined adjusted odds ratios for change in scores at T2 among the three PTG groups at T1 by combat deployment severity (see upper section of Table 3). Among those with moderate C-PTGI-SF scores at T1, service members who reported low or medium/high combat severity were significantly more likely to experience a decrease in their C-PTGI-SF scores at T2 compared with those who had no combat experiences (low combat intensity: OR = 1.33, 95% CI = 1.08–1.64; medium/high combat intensity: OR = 1.31, 95% CI = 1.04–1.66).

TABLE 3. Adjusted^a Odds of Experiencing a Positive or Negative Change in C-PTGI-SF Scores from T1 to T2, with a Reference Group of No Change

	Low PTG (n = 1320)		Moderate PTG (n = 5902)				High PTG (n = 1510)
	Positive Change (n = 369)		Negative Change (n = 663)		Positive Change (n = 526)		Negative Change (n = 348)
Characteristics	AOR	95% CI	AOR	95% CI	AOR	95% CI	AOR	95% CI
Combat Severity Status (ref: Deployed without combat)
Deployed, low combat	0.80	(0.59, 1.07)	1.33	(1.08, 1.64)	0.99	(0.80, 1.23)	0.96	(0.71, 1.29)
Deployed, medium/high combat	0.80	(0.56, 1.14)	1.31	(1.04, 1.66)	0.89	(0.69, 1.15)	0.95	(0.67, 1.35)
4-Level Mental Health Problems at Follow-Up (ref: No PTSD or depression)
Depression only	0.28	(0.09, 0.84)	5.35	(2.97, 9.64)	1.35	(0.51, 3.57)	9.02	(2.43, 33.55)
PTSD only	0.23	(0.10, 0.53)	6.56	(4.63, 9.29)	0.54	(0.22, 1.36)	3.96	(1.95, 8.05)
Comorbid PTSD and depression	0.15	(0.05, 0.44)	10.07	(6.41, 15.82)	–	–	11.33	(3.94, 32.54)

^aModels adjusted for the following variables: enrollment panel, sex, age, race/ethnicity, education level, marital status, military service branch, pay grade, service component, military occupation, combat deployment history prior to T1, mental health status at T2, problem drinking, smoking status, and social support. Adjusted odds ratios were not presented for the model examining positive change among those starting with moderate C-PTGI-SF scores at T1 due to a cell count of n = 1.

Third, we reported a negative unadjusted correlation between C-PTGI-SF sum score at T2 and the PCL-C sum score at T2 (r = −0.38, p < .001; Table 4). Multivariable analyses of PTSD in relation to change categories were conducted to address research question 2. Newly-reported PTSD was significantly associated with negative change for those in the T1 moderate or high groups (see lower section of Table 3). Those in the T1 low group who developed PTSD were less likely to have a positive change compared with those who did not develop any mental health problem (AOR = 0.23, 95% CI: 0.10–0.53).

TABLE 4. Correlations and Descriptive Statistics for C-PTGI-SF Scores and PTSD and Depression Scores

	C-PTGI-SF T1	C-PTGI-SF T2	PCL-C Sum T2	PHQ-8 Sum T2
Correlations
C-PTGI-SF sum T1	–	.59**	−.16**	−.20**
C-PTGI-SF sum T2	.59**	–	−.38**	−.41**
PCL-C sum T2	−.16**	−.38**	–	.77**
PHQ-8 sum T2	−.20**	−.41**	.77**	–
Descriptive Statistics
Mean	42.1	41.5	23.0	2.8
Standard Deviation	8.5	9.5	9.4	3.9
Range	0–55	0–55	17–85	0–24

Note: correlations with ** were significant at p < .0001; correlations with * were significant at p < .05. C-PTGI-SF sum T1 = Current standing, short form version of the Posttraumatic Growth Inventory sum score at time 1; C-PTGI-SF sum T2 = Current standing, short form version of the Posttraumatic Growth Inventory sum score at time 2; PCL-C sum T2 = PTSD Checklist, Civilian Version sum scores at time 2; PH8-sum T2 = Patient Health Questionnaire 8 item screening tool for depression sum score at time 2.

Fourth, we reported a negative unadjusted correlation between C-PTGI-SF sum score at T2 and PHQ-8 sum score at T2 (r = −0.41, p < .001; Table 4). Multivariable analyses of depression in relation to change categories were conducted to address research question 3. Newly-reported depression was significantly associated with negative change for those in the T1 moderate or high groups (see lower section of Table 3). Those in the T1 low group who developed depression were less likely to have a positive change compared with those who did not develop any mental health problem (AOR = 0.28, 95% CI: 0.09–0.84).

In terms of comorbidity, results were similar. Among those in the T1 moderate or high groups for PTG, those who developed comorbid PTSD and depression were over 10 times more likely to experience a negative change compared to those who did not develop a PTSD or depression. We note that few participants with moderate C-PTGI-SF scores at T1 who developed comorbid mental health problems experienced positive change, and so this estimate was not calculated due to small cell size (n = 1).

To further characterize PTG at T2, we examined data reported in Table 2. In terms of the 517 individuals who screened positive for mental health problems at T2, only 23 (4.4%) experienced any positive change in C-PTGI-SF scores at T2. In contrast, of the 8215 who did not screen positive for mental health problems at T2, 872 (10.6%) experienced positive change.

DISCUSSION

This study measured PTG longitudinally using an instrument that addressed psychometric problems that have plagued the PTG literature. Changes in C-PTGI-SF scores were examined over an approximate three-year period among service members before and after a combat deployment. The analysis was limited to service members who screened negative for PTSD or depression prior to deployment, so that any association between PTG and new reports of depression or PTSD could be assessed. The study found that positive changes in life perspective consistent with the construct of post-traumatic growth were inversely related to developing PTSD or depression following deployment. The single most important characteristic among individuals in this study who experienced positive change in perspective was their unwavering strong mental health over time; conversely, among those who developed PTSD or depression, less than 5% showed positive growth. Further, the negative correlations observed between these two mental health problems and PTG indicate that the development of mental health problems was associated with negative changes in life perspective. In addition, combat experiences, regardless of severity level, did not result in positive change in C-PTGI-SF scores for the vast majority of the sample. Specifically, after adjusting for the development of mental health problems, the traumatic experience of combat did not confer growth (positive changes in perspective over time) among 94% of all study participants. Taken together, these findings challenge the construct of PTG as it has been conceptualized and measured in the existing literature.

As described in the introduction, research findings examining the association between PTG and combat experience have been mixed (Mark et al., 2018). In our study, we observed that those with moderate C-PTGI-SF scores at T1 who reported low or medium/high levels of combat severity were more likely to have a negative change in C-PTGI-SF scores at T2, independent of mental health status at T2. This finding stands in partial contrast to the negative quadratic relationship (inverted u-shape) reported by McLean et al. (2013), where the PTGI scores rose as combat stress increased from low to moderate levels, but diminished at higher stress levels. Our results also differ from Bush et al. (2011), Gallaway et al. (2011), and Mitchell et al. (2013) who found a positive correlation between combat and PTGI scores, and Kaler et al. (2011) and Marotta-Walters et al. (2015) who found no correlation. It is also worth noting that participants in our study who had combat deployment(s) prior to T1 and entered the study with moderate C-PTGI-SF scores were more likely to experience a negative change in score; but this relationship was no longer significant after adjusting for mental health problems.

Regardless of whether these previous PTG and combat studies used cross-sectional or longitudinal designs, none measured actual growth, but relied on retrospective accounts. As noted by Frazier et al. (2009) and Park & Sinnott (2018), in studies with college students and cancer survivors, retrospective recall does not effectively capture actual changes in positive coping. More importantly, the direct linkage of the independent variable (trauma exposure) with the dependent variable (PTG) through the structure of the PTGI questions diminishes the scientific validity of correlations between these variables, a problem addressed by the present study.

In terms of the association between actual growth and mental health problems, the current study found PTG to be inversely related to both PTSD and depression. Other longitudinal studies have documented a positive association between PTSD and PTG, yet these studies are constrained by samples with mental health problems and retrospective measures of PTG (Dekel et al., 2012; Hall et al., 2015). In the present study, nearly every participant who developed PTSD and/or depression showed no positive change in C-PTGI-SF, and they were also significantly more likely to experience a negative change than those who did not develop a mental health problem. This pattern was evident regardless of participants’ C-PTGI-SF T1 score. Interestingly, even participants with high levels of C-PTGI-SF scores at T1 (i.e. a strong positive perspective) were 3 to 11 times more likely to experience a downward shift in their C-PTGI-SF scores by T2 if they developed a mental health problem, compared to those who did not develop a mental health problem. In addition, when analyses were replicated to include participants who screened positive for mental health problems at baseline, results remained nearly identical. These results suggest that a negative change in perspective and the development of a mental health problem may reflect similar psychological processes.

The present study offers a unique opportunity to examine both the association between mental health problems and C-PTGI-SF scores as well as the shift in these scores. Specifically, while the bivariate correlations between C-PTGI-SF at T2 and mental health problems were moderately correlated, the shift in C-PTGI-SF scores appear to be more strongly associated with mental health problems. This pattern suggests that it may be less important to consider the C-PTGI-SF score itself and more important to consider change, or the empirical shift in score over time.

Results documenting the relationship between actual growth and mental health problems in the present study may differ from previous studies because our methodology did not reflect asking individuals to attribute change in life perspective to the occurrence of a traumatic event. Previous studies assessing this kind of attribution process may involve “a self-deceptive strategy” for self-soothing during posttraumatic stress (Dekel et al., 2012). Such an explanation, coupled with our results, reinforces why it is necessary for studies of PTG to avoid attribution bias through truly independent measures, and so that actual growth is not confused with perceived growth or “illusory coping” (Zoellner & Maercker, 2006). Nevertheless, it is possible that considering components of the PTG construct (e.g., personal strength or appreciation of life) might help individuals reframe, cope with, and integrate traumatic experiences into one’s life perspective.

Despite the study’s unique strengths, there are limitations that should be considered. Floor effects and, to a greater extent, ceiling effects were evident in the current data. This attribute of the C-PTGI-SF scale in conjunction with groups identified by baseline C-PTGI-SF scores may lead to regression to the mean. Regression to the mean is reflected by the observations that positive change was more common in participants with low baseline C-PTGI-SF than moderate baseline C-PTGI-SF (28% vs 9%) while negative change was more common among participants with high baseline C-PTGI-SF than moderate baseline C-PTGI-SF (23% vs. 11%). However, regression to the mean is, by definition, a random process and so would not explain the consistency of our findings and the observed relationship between exposures of interest and change over time (Barnett et al., 2005). In terms of consistency, the main exposures of interest and several covariates were associated with changes C-PTGI-SF scores among those with low, moderate or high scores at T1, suggesting that changes were not just occurring among those regressing to the mean. Also the C-PTGI-SF questions on our survey have not been specifically validated for longitudinal assessment, although Kaur et al. (2017) reported sound psychometric properties in a cross-sectional examination of this measure. In addition, it is possible that some participants may have attempted to portray themselves more favorably through their responses on the C-PTGI-SF because this instrument did not identify a specific traumatic event, but rather measured specific traits self-reported by respondents. Also, “event-centrality” (Johnson & Boals, 2015), or the degree to which participants identified with combat trauma, was not measured in this study, but may be less relevant for veterans who have deployed to combat zones, as deployment is a hallmark occupational experience for those who have volunteered for military service. Furthermore, the average length of time between return home from deployment and the T2 assessment in this study was 1.3 years (SD = 0.9, range: 0–4); the follow-up time may not have been sufficient for PTG to occur for all participants, since research has suggested that 2 years is the optimal amount of time necessary for PTG to occur (Helgeson et al., 2006). Finally, this study does not address causality or direction of effect because mental health problems and C-PTGI-SF scores were measured simultaneously at T2.

Despite these limitations, this study provides a relatively non-biased look at PTG in comparison with other studies because 1) the lack of reference to a specific traumatic event eliminates retrospective attribution or recall biases, 2) changes in functioning between survey administration were directly measured longitudinally rather than relying on a retrospective self-assessment of change, 3) there was similarity in terms of type of trauma (combat experience), and 4) an average of more than 1 year passed between the traumatic event and follow-up, such that meaningful changes in perspective had sufficient time to develop.

Taken together, these results challenge the clinical utility of the PTG construct when measured empirically. PTG may be a useful cognitive framework on an individual level for coping with trauma and with mental health recovery (Johnson & Boals, 2015; Tsai et al., 2016), but from a clinical perspective, PTG does not appear to be distinct and independent from the negative psychological impact of traumatic experiences like combat. Findings from this study suggest that a patient’s change in positive outlook may be a useful way to structure discussion around coping with trauma. While the majority of individuals exposed to traumatic events do not develop mental health problems, the focus of clinical attention should remain on ensuring that those who develop mental health problems receive evidence-based trauma-focused treatment. An important remaining question is to what degree the construct of post-traumatic growth informs or overlaps with related domains, such as resilience, self-efficacy, and particularly the importance of expectations for recovery and healing. There is a considerable literature on the importance of facilitating positive expectations to enhance recovery and efficacy in mental health treatment. It has been suggested that the “personal strength” domain in the PTG construct is associated with lower incidence and severity of PTSD symptoms (Tsai et al., 2016), and treatment strategies that incorporate a growth or personal strength mind-set would be likely to facilitate expectation management (Constantino et al., 2018; Rief & Glombiewski, 2017). Additional longitudinal studies would help to sort this out, as well as studies that explore the construct of growth within the broader expectation, resilience, and self-efficacy literature. Better understanding of these factors could enhance therapeutic processes or help identify potential points for early intervention.

By refining measurement and methodology, understanding of the PTG construct can be expanded to consider an individual’s preexisting positive outlook, the importance of genuine shifts in outlook rather than an individual’s attributions regarding perceived change, and how reinforcing positive expectations and beliefs may support recovery from mental health problems following trauma.

AUTHORS’ CONTRIBUTION

IJ and AA drafted the introduction, results, and discussion; KR conducted the statistical analyses and drafted the methods; BP assisted with statistical analyses and interpretation of results; IJ supervised and assisted with the statistical analyses; CH assisted with drafting the abstract, introduction and discussion. All authors helped with designing the study, interpreted the results, provided critical analysis of the project, critically reviewed the manuscript, and approved of the final version.

DECLARATION OF INTERESTS

I am a military service member or employee of the U.S. Government. This work was prepared as part of my official duties. Report Number 20-85 was financially supported by the Military Operational Medicine Research Program, Defense Health Program, and Veterans Affairs under work unit no. 60002. The views expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Department of the Navy, Department of Defense, Uniformed Services University of the Health Sciences, nor the U.S. Government. The study protocol was approved by the Naval Health Research Center Institutional Review Board in compliance with all applicable Federal regulations governing the protection of human subjects. Research data were derived from an approved Naval Health Research Center, Institutional Review Board protocol number NHRC.2000.0007.

ACKNOWLEDGMENTS

In addition to the authors, Millennium Cohort Study team includes Richard Armenta, PhD; Satbir Boparai, MBA; Felicia Carey, PhD; Sheila Castañeda, PhD; Toni Rose Geronimo-Hara, MPH; Claire Kolaja, MPH; Rayna Matsuno, PhD; Deanne Millard; Chiping Nieh, PhD; Teresa Powell, MS; Anna Rivera, MPH; Beverly Sheppard; Daniel Trone, PhD; Jennifer Walstrom; and Steven Warner, MPH. The authors also appreciate contributions from the Deployment Health Research Department, Millennium Cohort Family Study, the Birth and Infant Health Research Team, and the Henry M. Jackson Foundation. We thank the Millennium Cohort Study participants.

SUPPLEMENTARY MATERIAL

Supplemental data for this article can be accessed on the publisher’s website.

Additional information

Funding

This work was supported by the Defense Health Program [n/a: program funds]; The Millennium Cohort Study is funded through the Military Operational Medicine Research Program, Defense Health Program, and Veterans Affairs [n/a: program funds].

Notes on contributors

Isabel G. Jacobson

Isabel G. Jacobson, MPH, is an epidemiologist with Leidos, Inc. at the Naval Health Research Center in San Diego, California. She has worked on the Millennium Cohort Study for more than 15 years, and has published 50 articles using data from this cohort. Her research focuses on behavioral health to promote well-being among service members, Veterans and their families.

Amy B. Adler

 Amy B. Adler, Ph.D. is a senior scientist with the Center for Military Psychiatry and Neuroscience at the Walter Reed Army Institute of Research. She has published more than 150 articles, co-edited 6 books, and is a Fellow of the American Psychological Association. Her current research focuses on small teams, behavioral health leadership, and rapid interventions to reduce acute stress.

Kimberly A. Roenfeldt

 Kimberly A. Roenfeldt, MAS, has a Master of Applied Statistics from Penn State University, and a Bachelor’s degree in Molecular, Cellular, and Developmental Biology from the University of Colorado. She worked as a Data Analyst for the Naval Health Research Center for two years, researching public health issues in the military. For the last two years, she has performed statistical programming and analysis in the pharmaceutical industry, specializing in clinical trials for therapies designed to treat central nervous system diseases.

Ben Porter

 Ben Porter, PhD, is an Assistant Research Professor at the Social Science Research Center at Mississippi State University. He was a Leidos contractor at the Naval Health Research Center from 2013-2021 and performed much of the work on the current project in that position. His research focuses on military health, particularly Gulf War Veteran health and trajectories of health overtime.

Cynthia A. LeardMann

 Cynthia A. LeardMann, MPH, is an epidemiologist at the Deployment Health Research Department, Naval Health Research Center, San Diego, California. Over the last 16 years, she has been an investigator with the Millennium Cohort Study, Millennium Cohort Family Study, and the Recruit Assessment Program. Her research focuses on psychological health and sexual trauma among US Service members, Veterans, and their Families. She earned her Master’s in Public Health from San Diego State University.

Rudolph P. Rull

 Rudolph P. Rull, PhD, MPH, is a Research Epidemiologist in the Deployment Health Research Department at the Naval Health Research Center and is the Principal Investigator of the Millennium Cohort Study. His research focuses on the long-term impacts of military service and deployments on the long-term health of US service members and veterans.

Charles W. Hoge

 Charles W. Hoge, M.D., Colonel (Ret), U.S. Army, is an internationally recognized expert on mental health in military service members and Veterans, and his articles on post-traumatic stress disorder (PTSD) and mild traumatic brain injury (mTBI) are some of the most cited medical articles from the wars in Iraq and Afghanistan. He is currently working as a Senior Scientist at Walter Reed Army Institute of Research and Office of the Army Surgeon General and a staff psychiatrist at Walter Reed National Military Medical Center.