Secondary Logo

Do post-traumatic pain and post-traumatic stress symptomatology mutually maintain each other? A systematic review of cross-lagged studies

Ravn, Sophie Lykkegaarda,b,c,*; Hartvigsen, Jand,e; Hansen, Maja; Sterling, Michelef; Andersen, Tonny Elmosea,b

doi: 10.1097/j.pain.0000000000001331
Systematic Review and Meta-Analysis
Global Year 2018
Global Year 2019

After traumatic exposure, individuals are at risk of developing symptoms of both pain and post-traumatic stress disorder (PTSD). Theory and research suggest a complex and potentially mutually maintaining relationship between these symptomatologies. However, findings are inconsistent and the applied methods are not always well suited for testing mutual maintenance. Cross-lagged designs can provide valuable insights into such temporal associations, but there is a need for a systematic review to assist clinicians and researchers in understanding the nature of the relationship. Thus, the aim of this systematic review was to identify, critically appraise, and synthesize results from cross-lagged studies on pain and PTSD symptomatology to assess the evidence for longitudinal reciprocity and potential mediators. Systematic searches resulted in 7 eligible studies that were deemed of acceptable quality with moderate risk of bias using the cohort study checklist from Scottish Intercollegiate Guidelines Network. Furthermore, synthesis of significant pathways in the cross-lagged models showed inconsistent evidence of both bidirectional and unidirectional interaction patterns between pain and PTSD symptomatology across time, hence not uniformly supporting the theoretical framework of mutual maintenance. In addition, the synthesis suggested that hyperarousal and intrusion symptoms may be of particular importance in these cross-lagged relationships, while there was inconclusive evidence of catastrophizing as a mediator. In conclusion, the findings suggest an entangled, but not necessarily mutually maintaining relationship between pain and PTSD symptomatology. However, major variations in findings and methodologies complicated synthesis, prompting careful interpretation and heightening the likelihood that future high-quality studies will change these conclusions.

aThRIVE, Department of Psychology, University of Southern Denmark, Odense, Denmark

bInCoRE, Department of Psychology, University of Southern Denmark, Odense, Denmark

cThe Specialized Hospital for Polio and Accident Victims, Roedovre, Denmark

dDepartment of Sports Science and Clinical Biomechanics, University of Southern Denmark, Odense, Denmark

eNordic Institute of Chiropractic and Clinical Biomechanics, Odense, Denmark

fRecover Injury Research Centre, NHMRC Centre of Research Excellence in Recovery Following Road Traffic Injuries, University of Queensland, Australia

Corresponding author. Address: Department of Psychology, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark. Tel.: +45 2681 9022. E-mail address: (S.L. Ravn).

Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (

Received April 13, 2018

Received in revised form June 15, 2018

Accepted June 26, 2018

Back to Top | Article Outline

1. Introduction

In recent years, there has been increased focus on the coexistence of pain and post-traumatic stress disorder (PTSD) symptomatology42 as well as on the nature of their potentially interacting relationship.7,10 According to DSM-IV,2 PTSD is a maladaptive reaction to traumatic exposure comprised 3 symptom clusters (ie, intrusion, avoidance, and hyperarousal), whereas PTSD as defined by DSM-53 also includes a fourth cluster of negative alterations in cognitions and mood. Also, the DSM-5 criteria constitute a more inclusive and heterogeneous condition than the DSM-IV.23 Although PTSD is not the only type of post-traumatic response, it is highly prevalent after traumatic exposure.33,34 According to DSM-5,3 traumatic exposure can be defined as exposure to actual or threatened death or serious injury, which a motor vehicle collision (MVC) or similar traumatic incidents can be examples of. In addition, DSM-IV2 also defined the exposure to incidents involving threat of physical integrity as a potential trauma (criterion A1), whereas there was also a demand for a response of intense fear, helplessness, or horror (criterion A2). Similarly, persistent pain is also common after a variety of traumatic injuries and events,9,15,28,43,49 making both post-traumatic stress symptoms and post-traumatic pain common after traumatic incidents.

Generally, studies report high rates of simultaneous pain and PTSD symptomatology both in pain populations and PTSD populations.42 Indeed, a recent meta-analysis of studies assessing PTSD symptomatology in pain samples reported a pooled mean prevalence of self-reported PTSD of 20.4%,54 indicating that a significant portion of patients with pain reports clinically relevant PTSD symptomatology. Post-traumatic stress disorder symptomatology is also associated with increased levels of pain, pain-related disability, and psychological distress across pain populations.1,24,46,48,50 Similarly, early levels of PTSD have been found to predict later pain and disability,13,28,30,35 and peritraumatic pain has also been found to be a risk factor for later PTSD symptoms.25,45 Indeed, this potential reciprocity of pain and PTSD symptomatology has been suggested in theoretical frameworks. Sharp and Harvey53 suggested that the 2 conditions mutually maintain one another through an array of cognitive, emotional, behavioural, and physiological factors. Similarly, Liedl and Knaevelsrud36 suggested that symptoms of pain and PTSD affect one another early after trauma and that symptoms of avoidance and hyperarousal directly interact with the pain experience. Asmundson et al.,5 on the other hand, suggested that the 2 conditions might share vulnerability factors such as anxiety sensitivity that could make an individual more prone to develop both conditions after trauma, whereas a recent systematic review also concluded that the 2 conditions share a number of neurobiological pathways that may also explain their interrelations.51

The view of mutual maintenance between symptoms of pain and PTSD has been widely applied and is also supported in recent nonsystematic literature reviews.7,10 However, most studies that have investigated the relationship between pain and PTSD were not designed to test for associations over time that could be interpreted as indicative of mutual maintenance. Instead, studies were primarily cross-sectional or only partially tested the suggested relationship, eg, PTSD symptoms as a predictor of later pain or vice versa. In addition, such associational patterns can in fact not be used to investigate mutual maintenance per se, as this is referring to a more complex set of processes than the mere testing of associations between pain and PTSD symptoms. This is complicated further by potential problems with separating symptoms of pain and PTSD, making assessments of PTSD symptoms within pain populations very likely to be inflated. Taken together, this problematizes the interpretations that can be made of studies claiming to present evidence of mutual maintenance between pain and PTSD symptoms. Despite all of this, the view of mutual maintenance has often uncritically been accepted and applied throughout the field. Regardless of this critique, bidirectional relationships and temporal precedence between pain and PTSD symptomatology are still important to better understand the true nature of the longitudinal relationship between the 2 constructs after trauma. Such associations can be tested using autoregressive cross-lagged models, which allow for complicated testing over time in a single model. The “autoregressive” component means that the construct is regressed on earlier measures of itself, hence capturing and controlling for the stability of the construct itself, whereas the “cross-lagged” component means that a construct is predicted by earlier measures of one or several other constructs.31,52 Such models are also known as cross-lagged panel analyses or cross-lagged path analyses and have recently been used to a wider extent to assess the reciprocal relationship of pain and PTSD symptomatology. However, no study has reviewed and synthesized the results of these studies, which will help shed light on the nature of the relationship between pain and PTSD symptomatology and potentially aid and affect the work of clinicians and researchers.

The aim of the present systematic review was to systematically identify, critically appraise, and synthesize research investigating the reciprocal associations between post-traumatic pain and PTSD symptomatology using cross-lagged panel models or the equivalent as well as potential mediators in these relationships.

Back to Top | Article Outline

2. Methods

This work was conducted and reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to create a best-evidence synthesis (the PRISMA checklist is available online as supplemental digital content at A protocol of the study was registered in the International Prospective Register of Systematic Reviews (PROSPERO) on July 1, 2017 (no. CRD42017071607).

Back to Top | Article Outline

2.1. Search strategy

The databases of PsycINFO, PubMed, Web of Science, EMBASE, Scopus, and PILOTS were searched for eligible studies on February 21, 2018. A thoroughly and broad search string was developed with search words combined using the Boolean Logic operators (AND and OR). During the preliminary work, pilot searches were performed to ensure inclusiveness and relevance of search terms and to identify relevant subject headings. The same search strategy was performed in all databases.

(PTSD OR PTSS OR PTS OR PSS OR ASD OR post-traumatic stress*

OR post-traumatic stress* OR post traumatic stress* OR acute stress*)




(longitudinal* OR prospective* OR cohort* OR observation* OR path analys* OR cross-lagged* OR structural equation mod* OR SEM)

All search words were checked for being registered either as subject headings or mesh terms. If this was the case, the search word was both included as this with auto explosion and as a free text word. Reference lists of eligible studies were also screened for additional references to ensure exhaustiveness.

Back to Top | Article Outline

2.2. Eligibility criteria

Studies had to fulfil the following eligibility criteria to be included in the present systematic review:

  •  (1) peer-reviewed articles,
  • (2) written in English, Danish, Swedish, or Norwegian,
  • (3) using an observational, prospective design with ≥3 measurement points after a traumatic event, which was not required to explicitly state fulfilment of criterion A,
  • (4) on adult samples aged 16 years or older, and
  • (5) applying autoregressive cross-lagged panel analysis or the equivalent.

Studies were excluded if they:

  •  (1) were intervention studies, reviews, dissertations, letters, editorials, book chapters, qualitative studies, cross-sectional studies, longitudinal studies not using a relevant design, and conference abstracts,
  • (2) did not include specific measurements of both pain intensity/severity and PTSD symptomatology,
  • (3) included samples where the index trauma was pregnancy and/or birth, or
  • (4) included participants with severe head injuries (<9 on the Glasgow Coma Scale), severe neurological diseases, and/or severe psychiatric comorbidities.
Back to Top | Article Outline

2.3. Data screening

First, all references were transferred to EndNote (version X8), and duplicates were removed. Both the removed and remaining references were double-checked manually. The remaining unique studies were then transferred to the online tool of Covidence, where the first and last author screened titles and abstracts independently according to the eligibility criteria. In cases of doubt or disagreement, the studies were carried forward for full-text screening. Here, disagreement between the raters was discussed among the raters themselves and with the inclusion of the second author, if needed. At this step, excluded papers and reasons for exclusion were logged into Covidence.

Back to Top | Article Outline

2.4. Data extraction

Two authors (first and last authors) independently extracted data from the included studies into a table for publication (Table 1). Again, discrepancies among the raters were discussed and corrected accordingly.

Table 1-a

Table 1-a

Table 1-b

Table 1-b

Back to Top | Article Outline

2.5. Assessment of methodological quality and risk of bias

The overall quality of the included studies was rated based on assessment of the methodological quality and the risk of bias. For this purpose, the checklist for cohort studies from Scottish Intercollegiate Guidelines Network (SIGN; available at was used with some modifications to fit our purpose. The modifications included deletion of some of the original items that were not applicable for the types of studies included in this study, which included original items 1.2, 1.3, 1.8, and 1.9, as well as item 1.12 (an exclusion criterion). In addition, 2 items regarding statistical assumptions and model fit were added, and the demonstration of validity and reliability of outcome assessments (original item 1.11) was split as 2 independent items based on the recommendation of Cancelliere et al.14 For every item, it was assessed whether the study in question appropriately did what was asked with a statement of “yes”, “no”, “can't say”, and sometimes “does not apply”. Ratings were made independently by the first and last authors, and afterwards compared and concurred. The second author was included in this process to ensure accuracy. Based on completed checklists and the discussions among the raters, each study was then rated as having little or no risk of bias, moderate risk of bias, or high risk of bias.

Back to Top | Article Outline

2.6. Data synthesis

First, descriptive characteristics of included studies were tabulated. In-depth synthesis of the findings in the studies was then performed in 2 steps. First, studies using PTSD symptomatology as one total severity variable and studies using PTSD symptom clusters were considered separately. Next, findings from the studies investigating mediators of the cross-lagged relationship between pain and PTSD symptomatology were synthesized. Because of differences in the specific measurement time points, studies were described in detail on this count in each synthesis to enhance visibility. For this purpose, significant cross-lagged associations were assessed based on the reported P values. When both cross-lagged coefficients were significant (P < 0.05), this was indicative of a bidirectional (ie, mutual) maintenance pattern. When only one cross-lagged coefficient was significant, this was indicative of a unidirectional maintenance pattern. Finally, when no cross-lagged coefficients were significant, this was indicative of no maintenance pattern. (Fig. 1).

Figure 1

Figure 1

Back to Top | Article Outline

3. Results

3.1. Identification of studies

Based on our search terms, a total of 7164 studies were identified across the databases. A total of 2838 duplicates were removed, leaving 4326 unique publications for screening. Of these, 97 were found eligible for full-text screening. A total of 7 studies matched our eligibility criteria, whereas the other 90 were excluded primarily based on study design. Two of the 7 studies were based on the same study sample, as Carty et al.16 used a subsample of the full sample used by Liedl et al.,37 which was confirmed by contacting the author group. Screenings of reference lists of the 7 included studies did not reveal additional studies (Fig. 2).

Figure 2

Figure 2

Back to Top | Article Outline

3.2. Descriptive characteristics

The 7 eligible studies included a total of 2773 unique participants, counting patients with severe injury without severe traumatic brain injury,16,29,37 patients with minor injury post-MVC,21,47 blast-exposed military personnel,56 and burn victims.60 Around one-third of the participants were females (35.92% across studies [range: 3.35%-66.40%]). Mean age ranged from 27.4 to 40.9 years in the 6 studies that reported this with a cross-study mean of 37.7 years. Descriptive characteristics for each of the 7 studies and their samples are presented in Table 1.

Back to Top | Article Outline

3.3. Risk of bias assessments

All studies were appraised to be of acceptable methodological quality with moderate risk of bias. Of note, this category contains great variability, as the amount and types of methodological issues vary from study to study, which is visualized in Table 2.

Table 2

Table 2

None of the studies included assessment of preinjury status of the outcomes, which introduces a risk of performance bias relating to the selection of subjects, as it was generally not ensured that the participants did not have either preinjury pain or PTSD symptomatology. This was, however, partly addressed by Feinberg et al.,21 who excluded participants with non–MVC-related axial pain above a certain threshold, and Ravn et al.,47 who excluded participants with previous whiplash-associated disorder. Most studies also had potential attrition bias with a significant subset of the sample dropping out over time,16,29,37,56,60 with only Feinberg et al.'s21 dropout rate not exceeding the recommended 20% and Ravn et al.47 not reporting dropout rates at all. Four studies applied dropout analyses.16,37,56,60 All included studies had clearly defined outcomes and used validated assessment tools of both PTSD and pain. Despite this, however, the outcomes were not consistently focused on trauma-related symptoms, and the studies often failed to refer to other studies assessing the psychometric properties of the scales, introducing potential detection biases. In addition, there may be a potential bias associated with self-report of traumatic exposure47,56 compared with studies sampling from hospitals,16,21,29,37,60 especially when there is a significant time gap between trauma and baseline assessment,56 introducing an additional risk of recall bias. Relatedly, none of the studies were explicitly clear about endorsement of criteria A1 and A2, something particularly important in studies when assessing PTSD symptomatology in samples exposed to objectively minor events.21,47 Furthermore, there may exist potential confounding-related concerns. Although all studies included the risk of confounding to some degree in designing the study and discussing the results, this was often only briefly touched on. Also, only 2 studies statistically controlled for demographics such as age and sex,47,56 and one controlled for catastrophizing by including it as a potential mediator.16 In addition, only Van Loey et al.60 provided confidence intervals for the path coefficients, and Feinberg et al.21 failed to report model fit indices, which is an issue in terms of assessing the legitimacy of the model. Finally, there was a general lack of commenting on the statistical assumptions, leaving the reader unable to judge potential biases related to this, with one study violating the assumption of stationarity by using different assessments of PTSD symptomatology at different time points.29

Back to Top | Article Outline

3.4. Synthesis of association patterns

Six studies investigated the cross-lagged relationship between pain and PTSD symptomatology, whereas the seventh study used fixed paths limited to associations with catastrophizing,60 hence only illuminating the relationship between pain and PTSD symptomatology through catastrophizing and not directly. Hence, only 6 studies are relevant for this section.

Four of the 6 studies investigated PTSD symptomatology as a total severity score. Of these, 3 reported evidence of bidirectional associations between pain and PTSD symptoms from T1 to T2,16,29,56 whereas this changed to unidirectional patterns from T2 to T3, either from pain to PTSD symptoms16 or from PTSD symptoms to pain.29,56 However, assessment points varied between studies with T2 being 316 and 6 months29 after injury, whereas the third study by Stratton et al.56 also had T2 at 6 months after baseline, but instead had a significantly longer mean period between trauma and baseline (mean 552 days). The fourth study found that only PTSD symptoms predicted pain from T1 (<4 weeks after injury) to 3 months after injury and again from 6 to 12 months after injury, whereas no relations were found from 3 to 6 months after injury.47

The remaining 2 of the 6 studies investigated PTSD symptom clusters (intrusion, hyperarousal, and avoidance), and both found evidence of bidirectional associations between hyperarousal and pain in the early months after trauma from T1 (less than 6 weeks after trauma) to T2 (either 3 or 6 months after trauma), and bidirectional associations between intrusion and pain in the chronic months after trauma from T2 (either 3 or 6 months after trauma) to T3 (12 months after trauma).21,37 In addition, Liedl et al.37 also found evidence of bidirectional associations between hyperarousal and pain from T2 to T3. Furthermore, a number of unidirectional effects were found with intrusion to pain37 and pain to intrusion21 found in early months after trauma (from T1 to T2), whereas pain to avoidance37 and pain to hyperarousal21 were found while in the chronic months after trauma (from T2 to T3).

Back to Top | Article Outline

3.5. Synthesis of evidence of mediators of association patterns

Two studies tested catastrophizing as a mediator in the models.16,60 Carty et al.16 found no evidence that catastrophizing was a mediator in the cross-lagged relationship between pain and PTSD symptoms, whereas Van Loey et al.60 found that PTSD symptoms at T1 predicted catastrophizing at T2, which then predicted pain at T3, indicating a mediating role of catastrophizing between initial PTSD symptoms and persistent pain at 12 months.

Back to Top | Article Outline

4. Discussion

The present systematic review identified 7 eligible studies, which were appraised to be of acceptable methodological quality with a moderate risk of bias related to possible performance, attrition, and detection biases as well as issues related to confounding and statistics. In synthesizing the findings of these studies, the present review found mixed evidence of both bidirectional and unidirectional associations between PTSD symptomatology and pain over time. Furthermore, the synthesis highlighted the importance of hyperarousal and intrusion symptoms in the cross-lagged relationship between pain and PTSD symptomatology, while there was inconclusive evidence of catastrophizing as a mediator between pain and PTSD symptomatology. In addition to the inconsistent findings across studies, the heterogeneity in study methodologies and the moderate risk of bias across all studies complicated synthesis. Hence, future high-quality studies may change these conclusions.

As our results did not uniformly confirm bidirectional association patterns between pain and PTSD symptomatology over time, which were used as indicative of potential mutual maintenance, our results only partly support the applied theoretical framework of mutual maintenance36,53 and the conclusions of existing nonsystematic reviews.7,10 However, the great variability in individual study findings obscures straightforward conclusions, for which there may be several contributing factors. One reason for cross-study discrepancies may be differences in trauma types or injury severities, eg, minor vs severe traumas, may be causing some sample types to display a more interconnected relationship between pain and PTSD symptomatology. However, we were not able to find any indication of such a pattern in the present review. In addition, differences in findings may be due to the different designs, as comparing findings from, for example, early post-trauma to 3 and 6 months, respectively, creates some important concerns, as both pain and PTSD symptomatology are fluctuating in nature.8,59 It may also be that there are certain time-determined differences in interactional patterns, but only very tentative patterns of this were found. Together, this may add further to the complexity in testing these cross-lagged relationships. Furthermore, the use of different assessment tools across studies may also capture both PTSD and pain symptomatology differently, thereby indirectly affecting the relationships tested. Indeed, a recent study showed that even very small changes in the wording in PTSD questionnaires changed the level of specific PTSD symptoms in patients with chronic pain,27 highlighting that even minor changes may change the interpretation and perhaps taps differently into the pain symptoms of the respondent. In addition, a part of the explanation may also be that the conditions influence each other indirectly through processes not captured by the present review such as, for example, elevated levels of (pain-related) distress.61,62 Finally, in terms of discussing the overall applicability of the theoretical viewpoint of mutual maintenance, it is important to note that the theory is likely to better apply in selected clinical samples with high levels of pain and PTSD, as it is possible that the reciprocity between the 2 constructs may be diluted when tested in more broad cohorts with varying (and generally very low) symptom levels.

Only 2 studies in the present review examined mediators in the cross-lagged models, both investigating the role of catastrophizing with divergent results.16,60 This difference may be due to design and statistical differences among the studies. The studies' second outcome assessment was at 316 and 660 months after trauma, while their statistical approach was also different, as Van Loey et al.60 did not include cross-lagged paths between pain and PTSD symptomatology. Similarly, only 2 studies examined the role of the individual PTSD symptom clusters,21,37 highlighting the importance of primarily hyperarousal and intrusion symptoms with both unidirectional and bidirectional effects over all time points. Avoidance symptoms were, on the other hand, only found to be of relevance at a single time point in one study,37 suggesting that avoidance behaviours are not central in the reciprocity of PTSD symptoms and pain. Of note, the DSM-IV avoidance symptom cluster, as used here, also contains numbing symptoms.2 Overall, studies on the relationships between pain and PTSD symptomatology have highlighted the centrality of especially hyperarousal symptoms13,18,32,38,40,55 and to a lesser degree intrusion.57 In addition, the importance of both clusters is highlighted in the theoretical perspectives of mutual maintenance.36,53 The importance of hyperarousal could rely on the tendency to catastrophic misperceptions and negative interpretations as well as anticipations of somatic sensations,32,36,53 which would therefore be predictive of pain, whereas intrusion is suggested to trigger pain and vice versa.53 It is, however, important to be critical in the interpretation of the findings regarding hyperarousal, as the finding that hyperarousal symptoms have a reciprocal relationship with pain may stem from the fact that hyperarousal symptoms are simply reflecting the pain experience itself. Research of the latent structure of PTSD has suggested that the hyperarousal clusters consist of both the so-called anxious arousal and dysphoric arousal,4 with the latter being more related to general distress and potentially pain-related symptomatology.

Several critical issues were identified in the risk of bias assessments with a few meriting further attention. First, one concern is related to the fact that none of the studies assessed preinjury symptomatology of pain and PTSD, which is likely to affect postinjury ratings and thereby cause skewed results. Second, the measurements of pain and PTSD were not consistently focused on a specific trauma exposure. Only Feinberg et al.21 and Jenewein et al.29 explicitly stated that the pain assessments were asking for accident-related pain, and only Ravn et al.47 and Van Loey et al.60 explicitly stated that the PTSD assessment was concerning MVC-related or burn-related PTSD symptoms, respectively. This forms a particular issue in the present review, as the relationship and relative influence of pain and PTSD symptomatology on each other may change heavily depending on whether or not the same trauma caused both conditions, hence undermining the interpretations that can be drawn. Third, there exists a potential validity issue of assessing PTSD symptomatology in participants with persistent pain, which stems from the fact that many PTSD symptoms included in the DSM criteria are not unique to this diagnosis.39 As such, PTSD responses may be inflated by pain-related symptomatology, thereby increasing the risk of false positives. At the same time, a number of other psychological conditions such as depression and anxiety, which are both very common in chronic pain samples,17 can add an additional risk of false-positive answers. Specifically, for the purpose of assessing PTSD symptomatology, studies using clinician-administered interviews consistently16,37 must be regarded of higher quality. A related validity concern is that the studies were generally not explicit about endorsement of criteria A1 and A2. Although the A2 criterion has been removed from the DSM-5, as it did not add to the predictive nor diagnostic value of PTSD,12,22 the potential lack of fulfilment of criterion A1 is something potentially very problematic. Particularly, this forms an potential issue in studies assessing PTSD symptomatology in samples experiencing minor injuries,21,47 as it is more likely that such objectively minor injuries and incidents may not fulfil the DSM-IV criteria on threat of death, serious injury, or physical integrity.2 As the criterion A1 is an important part of the diagnostic criteria, a lack of endorsement can indeed introduce a higher risk of validity biases in assessing PTSD symptomatology. However, we argue that objectively minor events can indeed be perceived as a threat of death, serious injury, and/or physical integrity, possibly more so in cases with neck traumas as compared to traumas in other parts of the body. In addition, even if criterion A1 is not endorsed for all, a recent study reported that the structural relations between PTSD symptoms were similar in patients who fulfill criterion A and patients who report a subthreshold stressor,64 suggesting that assessing PTSD symptomatology in samples with subthreshold stressors is still relevant. Despite this, however, it is still potentially critical in terms of interpretation and feeds into the debate of increased risk of false positives, prompting careful interpretations. A further point is that the majority of the studies in the present review16,29,37,47,56,60 also assessed PTSD symptomatology very early after trauma, whereas the DSM-IV diagnostic criterion is symptoms of at least 30 days to preclude normative transient responses.2 Hence, these assessments are very likely to capture a normative and transitory stress reaction that not necessarily has anything to do with later PTSD symptomatology, hence also challenging the validity of these assessments. Finally, the evaluation of cross-lagged associations relied on P values in all studies. As a P value is merely a measure for the probability of getting the present (or something beyond the present) result if the null hypothesis is indeed true,6 this is not a good indicator of clinical relevance.19 Instead, measures of the magnitudes of the associations (eg, a type of effect size) along with confidence intervals are a preferred way to assess the precision and relevance of the different associations. However, the majority of studies only presented (some of) this information for the significant associations and not the nonsignificant ones. In addition, these were standardized regression coefficients, which are problematic to compare in multivariable relationships, as they are then controlled for different variables across studies, making pooling of such effect sizes and their interpretation a challenge.44

The results of the present review have several implications. First of all, despite the findings underlining a close and potentially changing entanglement of the 2 conditions over time, it is important to not uncritically apply the mutual maintenance theory of PTSD symptomatology and pain. Furthermore, clinicians are encouraged to be aware of this complex relationship and how it may affect treatment outcomes. This implies that clinicians screen for both pain and PTSD symptoms after traumatic exposure and are attentive of any patterns of mutual maintenance, may be especially between pain and the PTSD symptom clusters of intrusion and hyperarousal. Future research should investigate the nature of the complex relationship between pain and PTSD symptoms with close attention to the methodological limitations addressed in the present review. Specifically, future studies should attempt to eliminate the risk of preinjury presence of the outcomes as well as the risk of false positives when assessing PTSD in patients with pain. Thus, we hope to encourage awareness of this potential issue and argue that future work on this should ensure endorsement of the criterion A and use clinically administered interviews or focus on the core symptoms of PTSD symptoms when using questionnaires,26,39,63 while neither, however, rule out the risk of false positives. In addition, future research should report on statistical assumptions, control and discuss the role of confounding, and include effect sizes along with confidence intervals.

Back to Top | Article Outline

4.1. Limitations

The present review is subjected to several limitations, which may influence the interpretation of the results. First, this study has pooled study findings regardless of the large methodological differences between studies, making interpretations of the findings more complicated and uncertain. Because of these large methodological variations across studies, it was not possible to undertake meta-analysis, which would have added significantly to a narrative synthesis.20 Although this would have been a stronger methodology, we do not think it would have had impact on our conclusions. Second, the risk of bias assessments was performed using a modified tool developed for observational cohort studies and not specifically cross-lagged modelling studies, which may give rise to risk of bias in the evaluation process by systematically evaluating the studies on potentially inadequate parameters. Third, the present review stated some exclusion criteria that the included studies did not report on, making the evaluation of eligibility unclear in some cases. Relatedly, a more explicitly trauma conceptualisation related to the diagnostic demand of the criterion A in DSM-IV2 would have allowed for stronger conclusions, as the lack of this poses a potential validity bias in the assessment of PTSD symptomatology. Fourth, autoregressive cross-lagged models have several limitations, which may bias the results. Among others, this technique assumes that all important predictors are in the model, something very hard to satisfy.31 Also, the technique assumes synchronicity, which holds that the constructs at a given time points are measured at exactly the same time, something often violated by practicalities in the data collection process.31 Fifth, the coefficients of the significant cross-lagged paths were generally small of size, indicating relatively weak associations, which was not taken into account in our analysis. Sixth, as PTSD in DSM-5 constitutes a more inclusive and heterogeneous condition compared with DSM-IV23 comprised 4 clusters and changes to criterion A,3 it may be that the findings in the present review may not be replicated in studies using DSM-5. Similarly, as post-traumatic distress varies across ethnic and cultural settings,11,58 the present findings may not be generalizable to other cultural settings. In addition, other types of post-traumatic distress than PTSD symptoms can be of relevance. Seventh, bidirectional associations were used as indicative of mutual maintenance across the studies and in the present review as well. However, mutual maintenance as a concept is much more holistic, process-oriented, and complex than the mere testing of reciprocal associations over time between 2 constructs, prompting critical interpretation and careful use of this terminology. Finally, a number of decisions made by the authors of this study may influence the results. For instance, a decision was made to ignore the baseline measurement in the Feinberg's study21 due to the fact that PTSD symptomatology was not measured at this time point. Also, there was no attempt to blind the assessors, which could potentially cause bias (especially since one study is conducted by the present author group47). Furthermore, the risk of bias assessments in the end, although strongly guided by the used tool, was subjective evaluations, causing all studies to deemed of acceptable level of quality despite major variations across them.

Back to Top | Article Outline

4.2. Conclusions and future directions

The findings of the present systematic review suggest an entangled relationship between pain and PTSD symptomatology over time after trauma with a potential importance of specifically hyperarousal and intrusion symptoms and may be also catastrophizing, however with major variations in the nature of this relationship across studies and time points. Therefore, these findings only partly and indirectly support the perspective of mutual maintenance between pain and PTSD symptomatology. In addition to difference in results across studies, synthesis was also complicated by large methodological differences between them as well as an increased risk of bias. All in all, these variations across findings as well as methodology are indicative of tentative findings, hence underlining the importance of very critical and careful interpretation. Hence, future high-quality studies may change these conclusions. Such future studies ought to minimize the risk of biases and the general limitations identified by the present review and potentially apply different methodologies. This may, among others, be ecological momentary assessments and qualitative approaches that can further clarify the nature and complexities of the relationship between pain and PTSD symptomatology by adding more detailed and process-related insights.

Back to Top | Article Outline

Conflict of interest statement

The authors have no conflict of interest to declare.

Back to Top | Article Outline


The authors thank the help of the research library in giving feedback on the searches.

Back to Top | Article Outline

Supplemental digital content

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

Back to Top | Article Outline


[1]. Aakerblom S, Perrin S, Rivano Fischer M, McCracken LM. The impact of PTSD on functioning in patients seeking treatment for chronic pain and validation of the post-traumatic diagnostic scale. Int J Behav Med 2017;24:249–59.
[2]. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 4th ed, text rev. Washington: American Psychiatric Association, 2000.
[3]. American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 5th ed. Washington: American Psychiatric Association, 2013.
[4]. Armour C, Müllerová J, Elhai JD. A systematic literature review of PTSD's latent structure in the diagnostic and statistical manual of mental disorders: DSM-IV to DSM-5. Clin Psychol Rev 2016;44:60–74.
[5]. Asmundson GJ, Coons MJ, Taylor S, Katz J. PTSD and the experience of pain: research and clinical implications of shared vulnerability and mutual maintenance models. Can J Psychiatry 2002;47:930–7.
[6]. Baker M. Statisticians issue warning over misuse of P values: policy statement aims to halt missteps in the quest for certainty. Nature 2016;531:151.
[7]. Beck JG, Clapp JD. A different kind of comorbidity: understanding post-traumatic stress disorder and chronic pain. Psychol Trauma 2011;3:101–8.
[8]. Bonanno GA, Mancini AD. Beyond resilience and PTSD: mapping the heterogeneity of responses to potential trauma. Psychol Trauma 2012;4:74–83.
[9]. Bortsov AV, Platts-Mills TF, Peak DA, Jones JS, Swor RA, Domeier RM, Lee DC, Rathlev NK, Hendry PL, Fillingim RB, McLean SA. Pain distribution and predictors of widespread pain in the immediate aftermath of motor vehicle collision. Eur J Pain 2013;17:1243–51.
[10]. Brennstuhl MTC, Montel S. Chronic pain and PTSD: evolving views on their comorbidity. Perspect Psychiatr Care 2015;51:295–304.
[11]. Brewin CR, Andrews B, Valentine JD. Meta-analysis of risk factors for post-traumatic stress disorder in trauma-exposed adults. J Consult Clin Psychol 2000;68:748–66.
[12]. Brewin CR, Lanius RA, Novac A, Schnyder U, Galea S. Reformulating PTSD for DSM-V: life after criterion A. J Trauma Stress 2009;22:366–73.
[13]. Buitenhuis J, de Jong PJ, Jaspers JPC, Groothoff JW. Relationship between post-traumatic stress disorder symptoms and the course of whiplash complaints. J Psychosom Res 2006;61:681–9.
[14]. Cancelliere C, Cassidy JD, Cote P, Hincapie CA, Hartivigsen J, Marras C, Boyle E, Kristman V, Hung R, Stålnacke BM, Rumney P, Coronado V, Holm LW, Borg J, Nygren-de Boussard C, Af Geijerstam JL, Keightley M. Protocol for a systematic review of prognosis after mild traumatic brain injury: an update of the WHO Collaborating Centre Task Force findings. Syst Rev 2012;1:17.
[15]. Carroll LJ, Holm LW, Hogg-Johnson S, Cote P, Cassidy JD, Haldeman S, Nordin M, Hurwitz EL, Carragee EJ, van der Velde G, Peloso PM, Guzman J. Course and prognostic factors for neck pain in whiplash-associated disorders (WAD): results of the bone and joint decade 2000–2010 task force on neck pain and its associated disorders. Spine 2008;33:83–92.
[16]. Carty J, O'Donnell M, Evans L, Kazantzis N, Creamer M. Predicting post-traumatic stress disorder symptoms and pain intensity following severe injury: the role of catastrophizing. Eur J Psychotraumatol 2011;2:5652.
[17]. Castro M, Kraychete D, Daltro C, Lopes J, Menezes R, Oliveira I. Comorbid anxiety and depression disorders in patients with chronic pain. Arq Neuropsiquiatr 2009;67:982–5.
[18]. Cho SK, Heiby EM, McCracken LM, Moon DE, Lee JH. Daily functioning in chronic pain: study of structural relations with post-traumatic stress disorder symptoms, pain intensity, and pain avoidance. Korean J Pain 2011;24:13–21.
[19]. Dahiru T. P-value, a true test of statistical significance? A cautionary note. Ann Ib Postgrad Med 2008;6:21–6.
[20]. Fagard RH, Staessen JA, Thijs L. Advantages and disadvantages of the meta-analysis approach. J Hypertens Suppl 1996;14:9–12.
[21]. Feinberg RK, Hu J, Weaver MA, Fillingim RB, Swor RA, Peak DA, Jones JS, Rathlev NK, Lee DC, Domeier RM, Hendry PL, Liberzon I, McLean SA. Stress-related psychological symptoms contribute to axial pain persistence after motor vehicle collision: path analysis results from a prospective longitudinal study. PAIN 2017;158:682–90.
[22]. Friedman MJ. Finalizing PTSD in DSM-5: getting here from there and where to go next. J Trauma Stress 2013;26:5.
[23]. Galatzer-Levy IR, Brewin RA. 636,120 ways to have post-traumatic stress disorder. Perspect Psychol Sci 2013;8:651–62.
[24]. Geisser ME, Roth RS, Bachman JE, Eckert TA. The relationship between symptoms of post-traumatic stress disorder and pain, affective disturbance and disability among patients with accident and non-accident related pain. PAIN 1996;66:207–14.
[25]. Glynn SM, Asarnow JR, Asarnow R, Shetty V, Elliot-Brown K, Black E, Belin TR. The development of acute post-traumatic stress disorder after orofacial injury: a prospective study in a large urban hospital. J Oral Maxillofac Surg 2003;61:785–92.
[26]. Hansen M, Hyland P, Armour C, Shevlin M, Elklit A. Less is more? Assessing the validity of the ICD-11 model of PTSD across multiple trauma samples. Eur J Psychotraumatol 2015;6:28766.
[27]. Hansen M, Hyland P, Karstoft KI, Vaegter HB, Bramsen RH, Nielsen ABS, Armour C, Andersen SB, Høybye MT, Larsen SK, Andersen TE. Does size really matter? A multisite study assessing the latent structure of the proposed ICD-11 and DSM-5 diagnostic criteria for PTSD. Eur J Psychotraumatol 2017;8:1398002.
[28]. Jenewein J, Moergeli H, Wittmann L, Buchi S, Kraemer B, Schnyder U. Development of chronic pain following severe accidental injury. Results of a 3-year follow-up study. J Psychosom Res 2009;66:119–26.
[29]. Jenewein J, Wittmann L, Moergeli H, Creutzig J, Schnyder U. Mutual influence of post-traumatic stress disorder symptoms and chronic pain among injured accident survivors: a longitudinal study. J Trauma Stress 2009;22:540–8.
[30]. Katz J, Asmundson GJ, McRae K, Halket E. Emotional numbing and pain intensity predict the development of pain disability up to one year after lateral thoracotomy. Eur J Pain 2009;13:870–8.
[31]. Kearney MW. Cross-lagged panel analysis. In: Allen MR, editor. Sage encyclopedia of communication research methods. Thousand Oaks: Sage Publications, 2017.
[32]. Kenardy J, Dunne R. Traumatic injury and traumatic stress. Spine 2011;36:233–7.
[33]. Kessler RC, Wai TC, Demler O, Walters EE. Prevalence, severity, and comorbidity of twelve-months DSM-IV disorders in the national comorbidity survey replication (NCS-R). Arch Gen Psychiatry 2005;62:617–27.
[34]. Kilpatrick DG, Resnick HS, Milanak ME, Miller MW, Keyes KM, Friedman MJ. National estimates of exposure to traumatic events and PTSD prevalence using DSM-IV and DSM-5 criteria. J Trauma Stress 2013;26:537–47.
[35]. Kongsted A, Bendix T, Qerama E, Kasch H, Bach FW, Korsholm L, Jensen TS. Acute stress response and recovery after whiplash injuries. A one-year prospective study. Eur J Pain 2008;12:455–63.
[36]. Liedl A, Knaevelsrud C. Chronic pain and PTSD: the perpetual avoidance model and its treatment implications. Torture 2008;18:69–76.
[37]. Liedl A, O'Donnell M, Creamer M, Silove D, McFarlane A, Knaevelsrud C, Bryant RA. Support for the mutual maintenance of pain and post-traumatic stress disorder symptoms. Psychol Med 2010;40:1215–23.
[38]. Lopez-Martinez AE, Ramirez-Maestre C, Esteve R. An examination of the structural link between post-traumatic stress symptoms and chronic pain in the framework of fear-avoidance models. Eur J Pain 2014;18:1129–38.
[39]. Maercker A, Brewin CR, Bryant RA, Cloitre M, Reed GM, van Ommeren M, Humayun A, Jones LM, Kagee A, Llosa AE, Rousseau C, Somasundaram DJ, Souza R, Suzuki Y, Weissbecker I, Wessely SC, First MB, Saxena S. Proposals for mental disorders specifically associated with stress in the International Classification of Diseases-11. Lancet 2013;381:1683–5.
[40]. Maujean A, Gullo MJ, Andersen TE, Ravn SL, Sterling M. Post-traumatic stress symptom clusters in acute whiplash associated disorder and their prediction of chronic pain-related disability. Pain Rep 2017;2:e631.
[41]. Moher D, Liberati A, Tetzlaff J, Altman DG; The PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 2009;6:e1000097. doi: .
    [42]. Moeller-Bertram T, Keltner J, Strigo IA. Pain and post traumatic stress disorder—review of clinical and experimental evidence. Neuropharmacology 2012;62:586–97.
    [43]. Nampiaparampil DE. Prevalence of chronic pain after traumatic brain injury: a systematic review. JAMA 2008;300:711–9.
    [44]. Nieminen P, Lehtiniemi H, Vähäkangas K, Huusko A, Rautio A. Standardised regression coefficient as an effect size index in summarising findings in epidemiological studies. Epidemiol Biostats Public Health 2013;20:e8854–15.
    [45]. Norman SB, Stein MB, Dimsdale JE, Hoyt DB. Pain in the aftermath of trauma is a risk factor for post-traumatic stress disorder. Psychol Med 2007;38:533–42.
    [46]. Pedler A, Sterling M. Patients with chronic whiplash can be subgrouped on the basis of symptoms of sensory hypersensitivity and post-traumatic stress. PAIN 2013;154:1640–8.
    [47]. Ravn SL, Sterling M, Lahav Y, Andersen TE. Reciprocal associations of pain and post-traumatic stress symptoms after whiplash injury: a longitudinal, cross-lagged study. Eur J Pain 2018;22:926–34.
    [48]. Ravn SL, Vaegter HB, Cardel T, Andersen TE. The role of post-traumatic stress symptoms on chronic pain outcomes in chronic pain patients referred to rehabilitation. J Pain Res 2017;11:527–36.
    [49]. Rosenbloom BN, Khan S, McCartney C, Katz J. Systematic review of persistent pain and psychological outcomes following traumatic musculoskeletal injury. J Pain Res 2013;6:39–51.
    [50]. Ruiz-Parraga GT, Lopez-Martinez AE. The contribution of post-traumatic stress symptoms to chronic pain adjustment. Health Psychol 2014;33:958–67.
    [51]. Scioli-Salter ER, Forman DE, Otis JD, Gregor K, Valovski I, Rasmusson AM. The shared neuroanatomy and neurobiology of comorbid chronic pain and PTSD: therapeutic implications. Clin J Pain 2015;31:363–74.
    [52]. Selig JP, Little TD. Autoregressive and cross-lagged panel analysis for longitudinal data. In: Laursen B, Little TD, Card NA, editors. Handbook of developmental research methods. New York: The Guilford Press, 2012.
    [53]. Sharp TJ, Harvey AG. Chronic pain and post-traumatic stress disorder: mutual maintenance? Clin Psychol Rev 2001;21:857–77.
    [54]. Siqveland J, Hussain A, Lindstrom JC, Ruud T, Hauff E. Prevalence of post-traumatic stress disorder in persons with chronic pain: a meta-analysis. Front Psychiatry 2017;8:164.
    [55]. Sterling M, Chadwick BJ. Psychologic processes in daily life with chronic whiplash: relations of post-traumatic stress symptoms and fear-of-pain to hourly pain and uptime. Clin J Pain 2010;26:573–82.
    [56]. Stratton KJ, Clark SL, Hawn SE, Amstadter AB, Cifu DX, Walker WC. Longitudinal interactions of pain and post-traumatic stress disorder symptoms in U.S. Military service members following blast exposure. J Pain 2014;15:1023–32.
    [57]. Taylor B, Carswell K, Williams AC. The interaction of persistent pain and post-traumatic re-experiencing: a qualitative study in torture survivors. J Pain Symptom Manage 2013;46:546–55.
    [58]. Trepasso-Grullon E. Differences among ethnic groups in trauma type and PTSD symptom severity. GSJP 2012;4.
    [59]. Trevino CM, deRoon-Cassini T, Szabo A, Brasel K. Acute longitudinal pain trajectories in the traumatically injured. SOJ Anesthesiol Pain Manag 2015;2:1–7.
    [60]. Van Loey NE, Klein-König I, de Jong AEE, Hofland HWC, Vandermeulen E, Engelhard IM. Catastrophizing, pain and traumatic stress symptoms following burns: a prospective study. Eur J Pain 2018;22:1151–1159.
    [61]. Vlaeyen JW, Crombez G, Linton SJ. The fear-avoidance model of pain. PAIN 2016;157:1588–9.
    [62]. Vlaeyen JW, Linton SJ. Fear-avoidance and its consequences in chronic musculoskeletal pain: a state of the art. PAIN 2000;85:317–32.
    [63]. Walton JL, Cuccurullo LAJ, Raines AM, Vidaurri DN, Allan NP, Maieritsch KP, Franklin CL. Sometimes less is more: establishing the core symptoms of PTSD. J Trauma Stress 2017;30:254–8.
    [64]. Zelazny K, Simms LJ. Confirmatory factor analyses of DSM-5 post-traumatic stress disorder symptoms in psychiatric samples differing in criterion A status. J Anxiety Disord 2015;34:15–23.

    Pain; Post-traumatic stress; PTSD; Systematic review; Autoregressive cross-lagged models; SEM

    Supplemental Digital Content

    Back to Top | Article Outline
    © 2018 International Association for the Study of Pain