Skip to main content Accessibility help
×
Home
Hostname: page-component-684899dbb8-ct24h Total loading time: 0.68 Render date: 2022-05-24T03:27:47.315Z Has data issue: true Feature Flags: { "shouldUseShareProductTool": true, "shouldUseHypothesis": true, "isUnsiloEnabled": true, "useRatesEcommerce": false, "useNewApi": true }

Female sex and femininity independently associate with common somatic symptom trajectories

Published online by Cambridge University Press:  10 November 2020

Aranka V. Ballering*
Affiliation:
University of Groningen, University Medical Center of Groningen, Interdisciplinary Center Psychopathology and Emotion Regulation (ICPE), P.O. Box 30.001, 9700 RB, Groningen, The Netherlands
Klaas J. Wardenaar
Affiliation:
University of Groningen, University Medical Center of Groningen, Interdisciplinary Center Psychopathology and Emotion Regulation (ICPE), P.O. Box 30.001, 9700 RB, Groningen, The Netherlands
Tim C. olde Hartman
Affiliation:
Department of Primary and Community Care, Radboud University Medical Center, P.O. Box 9101, 6500 HB, Nijmegen, The Netherlands
Judith G. M. Rosmalen
Affiliation:
University of Groningen, University Medical Center of Groningen, Interdisciplinary Center Psychopathology and Emotion Regulation (ICPE), P.O. Box 30.001, 9700 RB, Groningen, The Netherlands
*
Author for correspondence: Aranka V. Ballering, E-mail: a.v.ballering@umcg.nl
Rights & Permissions[Opens in a new window]

Abstract

Background

Multiple predictors have been associated with persistent somatic symptoms. However, previous studies problematically defined the persistence of symptoms, conflated participants' sex and gender, and focused on patient populations. Therefore, we studied associations between predictors, especially sex and gender, and longitudinal patterns of somatic symptoms in the general adult population. We also assessed whether predictors for persisting symptoms differ between sexes.

Method

To identify developmental trajectories of somatic symptoms, assessed by the SCL-90 SOM, we used latent class trajectory modeling in the Dutch Lifelines Cohort Study [N = 150 494; 58.6% female; median time to follow-up: 46.0 (min–max: 22.0–123.0) months]. To identify predictors of trajectories, we applied multiple logistic regression analyses. Predictors were measured by surveys at baseline and a composite gender index was previously developed.

Results

A five-class linear LCGA model fitted the data best: 93.7% of the population had a stable symptom trajectory, whereas 1.5% and 4.8% of the population had a consistently increasing or decreasing symptom trajectory, respectively. Female sex predicted severe, stable symptom severity (OR 1.74, 95% CI 1.36–2.22), but not increasing symptom severity (OR 1.15, 95% CI 0.99–1.40). Femininity was protective hereof (OR 0.60, 95% CI 0.44–0.82 and OR 0.66, 95% CI 0.51–0.85, respectively). Merely a few predictors of symptom severity, for instance hours of paid employment and physical functioning, differed in strength between sexes. Yet, effect sizes were small.

Conclusion

Female sex and femininity predict symptom trajectories. No large sex differences in the strength of additional predictors were found, thus it may not be clinically useful to distinguish between predictors specific to male or female patients of persistent somatic symptoms.

Type
Original Articles
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited.
Copyright
Copyright © The Author(s) 2020. Published by Cambridge University Press

Introduction

A substantial proportion of the general practitioner (GP) visits in the Netherlands (13–43%) are related to common somatic symptoms for which no sufficient cause can be found after adequate physical examinations and interventions (Olde Hartman et al., Reference Olde Hartman, Blankenstein, Molenaar, Bentz van den Berg, Van der Horst, Arnold and Woutersen-Koch2013; Verhaak, Meijer, Visser, & Wolters, Reference Verhaak, Meijer, Visser and Wolters2006). Persistence of these somatic symptoms is associated with increased functional impairment, feelings of internalized stigma and social isolation (Dirkzwager & Verhaak, Reference Dirkzwager and Verhaak2007). In addition to personal hardship, persistent common somatic symptoms may pose an economic burden on both an individual and societal level (Joustra, Janssens, Bültmann, & Rosmalen, Reference Joustra, Janssens, Bültmann and Rosmalen2015; Konnopka et al., Reference Konnopka, Schaefert, Heinrich, Kaufmann, Luppa, Herzog and König2012).

A systematic review from 2009 showed that 10–30% of patients with medically unexplained somatic symptoms attending the GP or secondary care clinic did not improve during their follow-up period of 6–15 months (olde Hartman et al., Reference Olde Hartman, Borghuis, Lucassen, van de Laar, Speckens and van Weel2009). More recent studies suggest higher rates of non-remission in primary care with 0.5 year (55.1%) (Lamahewa, Buszewicz, Walters, Marston, & Nazareth, Reference Lamahewa, Buszewicz, Walters, Marston and Nazareth2019), 1 year (51.2%) (Steinbrecher & Hiller, Reference Steinbrecher and Hiller2011), and 2 years follow-up (56.8% and 37.1%) (Budtz-Lilly, Vestergaard, Fink, Carlsen, & Rosendal, Reference Budtz-Lilly, Vestergaard, Fink, Carlsen and Rosendal2015; Claassen-van Dessel, van der Wouden, Hoekstra, Dekker, & van der Horst, Reference Claassen-van Dessel, van der Wouden, Hoekstra, Dekker and van der Horst2018). However, studies describing the persistence of common somatic symptoms are difficult to compare due to methodological differences. Furthermore, these figures might not be representative of the general adult population. The prognosis of common somatic symptoms in the general population is likely to be more favorable, since by definition patient populations suffer from symptoms that they regard serious enough to visit a physician. To the best of our knowledge, only one recent study on the persistence of unexplained common somatic symptoms has been conducted in the adult general population, indicating that 36.4% of people had persistent common somatic symptoms measured over 3 years (van Eck van der Sluijs et al., Reference van Eck van der Sluijs, ten Have, de Graaf, Rijnders, van Marwijk and van der Feltz-Cornelis2018).

A variety of predictors for persisting common somatic symptoms has been identified in adolescent and adult populations, including female sex (Janssens, Klis, Kingma, Oldehinkel, & Rosmalen, Reference Janssens, Klis, Kingma, Oldehinkel and Rosmalen2014; Steinbrecher & Hiller, Reference Steinbrecher and Hiller2011), physical and psychiatric comorbidities (olde Hartman et al., Reference Olde Hartman, Borghuis, Lucassen, van de Laar, Speckens and van Weel2009), symptom characteristics (such as the duration, severity and heterogeneity) (Budtz-Lilly et al., Reference Budtz-Lilly, Vestergaard, Fink, Carlsen and Rosendal2015; Kooiman, Bolk, Rooijmans, & Trijsburg, Reference Kooiman, Bolk, Rooijmans and Trijsburg2004), and psychological traits (such as neuroticism, perfectionism, and health perceptions) (Bonvanie, Rosmalen, van Rhede van der Kloot, Oldehinkel, & Janssens, Reference Bonvanie, Rosmalen, van Rhede van der Kloot, Oldehinkel and Janssens2015; De Gucht, Fischler, & Heiser, Reference De Gucht, Fischler and Heiser2004; Janssens et al., Reference Janssens, Klis, Kingma, Oldehinkel and Rosmalen2014). Identification of predictors for persisting somatic symptoms is pivotal, as it allows for early detection, diagnosis, and treatment and it could provide concrete starting points for interventions aiming to reduce or prevent such symptoms as well (olde Hartman et al., Reference Olde Hartman, Borghuis, Lucassen, van de Laar, Speckens and van Weel2009). However, the definition of the persistence of symptoms in most aforementioned studies aiming to identify predictors was suboptimal. For example, persistence was defined based on an arbitrary number of contacts with the GP (Verhaak et al., Reference Verhaak, Meijer, Visser and Wolters2006). This does not distinguish between one's symptoms and healthcare-seeking behavior, which is especially problematic given the observation that patients who do not return to the GP often still experience symptoms (Koch et al., Reference Koch, van Bokhoven, Bindels, van der Weijden, Dinant and ter Riet2009). Furthermore, it remains unknown whether these predictors differ between women and men.

Recently, a cross-sectional study found that not only female sex, but also feminine gender, which encompasses the roles, behaviors, identities, and relationships of women prescribed by societal norms in a given context (Johnson, Greaves, & Repta, Reference Johnson, Greaves and Repta2009), is associated with the severity of common somatic symptoms and prevalence of chronic diseases (Ballering, Bonvanie, Olde Hartman, Monden, & Rosmalen, Reference Ballering, Bonvanie, Olde Hartman, Monden and Rosmalen2020). Gender, and its embodiment, is more dynamic than one's biological sex. Yet, gender and sex are often conflated in research. Therefore, to date, it remains unknown whether sex and gender independently impact the severity and persistence of somatic symptoms in the general adult population. In addition, most studies assessing the severity and persistence of somatic symptoms have considered the cohort under study as a homogeneous population, while the longitudinal patterns of symptom development may show significant heterogeneity in their directions (Claassen-van Dessel et al., Reference Claassen-van Dessel, van der Wouden, Hoekstra, Dekker and van der Horst2018). This means that in most studies variable patterns of somatic symptoms over time remain undetected.

We present the first large epidemiological cohort study to identify the predictors of longitudinal patterns of somatic symptoms in the general adult population, with a special emphasis on sex and gender differences. First, we will use latent class trajectory modeling to identify developmental patterns of symptom severity. Second, we will assess which predictors are associated with different trajectories. Third, we aim to study whether identified predictors for the persistence of common somatic symptoms differ between females and males. We hypothesize that female sex and femininity associate with increased severity of common somatic symptoms.

Methods

Setting

This study is based on data collected within the Dutch Lifelines Cohort Study. The Lifelines Cohort Study is a multi-disciplinary prospective population-based cohort study examining in a unique three-generation design the health and health-related behaviors of 167 729 persons living in the North of The Netherlands. It employs a broad range of investigative procedures in assessing the biomedical, socio-demographic, behavioral, physical, and psychological factors, which contribute to the health and disease of the general population, with a special focus on multi-morbidity and complex genetics. Extensive information on the cohort, design considerations, and recruitment procedures is provided elsewhere (Klijs et al., Reference Klijs, Scholtens, Mandemakers, Snieder, Stolk and Smidt2015; Scholtens et al., Reference Scholtens, Smidt, Swertz, Bakker, Dotinga, Vonk and Stolk2015). The Lifelines Cohort Study is performed according to the principles of the Declaration of Helsinki and in accordance with the UMCG's research code. The Lifelines Cohort Study is approved by the Medical Ethical Committee of the University Medical Center Groningen, The Netherlands (Scholtens et al., Reference Scholtens, Smidt, Swertz, Bakker, Dotinga, Vonk and Stolk2015). For the current study, we adhered to the STROBE statement and GRoLTS guidelines for reporting of our findings (Van De Schoot, Sijbrandij, Winter, Depaoli, & Vermunt, Reference Van De Schoot, Sijbrandij, Winter, Depaoli and Vermunt2017; von Elm et al., Reference von Elm, Altman, Egger, Pocock, Gøtzsche and Vandenbroucke2014). To conform to the SAGER guidelines we reported our findings stratified by sex (Heidari, Babor, De Castro, Tort, & Curno, Reference Heidari, Babor, De Castro, Tort and Curno2016).

Participants

Participants completed questionnaires on multiple topics including, but not limited to, demographics, health, personality, psychological and somatic symptoms, and psychosocial characteristics. These questionnaires asked for participants' biological sex. Hence, we refer to the participants as male and female, whereas we refer to masculinity and femininity when discussing gender.

In the current study, we used data from the adult participants gathered at four time points: at baseline [n = 148 643; mean age 44.2 years (s.d. = 12.8); 58.6% females], at a first follow-up time point [n = 124 443; mean age 46.5 years (s.d. = 12.8); 59.3% female; median time to follow-up: 13.0 (10–93) months], after a second follow-up time point [n = 95 137; mean age 48.1 years (s.d. = 12.8); 59.8% female; median time to follow-up: 25.0 (22–92) months], and after a third follow-up time point [n = 90 077; mean age 49.8 years (s.d. = 12.6); 59.1% females; median time to follow-up: 46.0 (22–123) months]. A more detailed overview of the population included is provided in online Supplementary Appendix A. Attrition rates after 1.5, 3, and 4 years were 16.8%, 36.2%, and 39.6%, respectively, compared to baseline. We did not find any indication for relevant systematic attrition: no meaningful associations between potential predictors of the severity of common somatic symptoms and attrition rates were found.

Variables

We assessed common somatic symptom severity by means of the 12-item ordinal Symptom CheckList-90 Somatization subscale (SCL-90 SOM; online Supplementary Appendix B). The 12 items refer to how much bother or distress someone experienced the past week due to somatic symptoms. This scale has been recommended for large-scale studies and has been shown to have sufficient measurement invariance over time, which makes it suitable to assess the measured concepts repeatedly over time (Rytilä-Manninen et al., Reference Rytilä-Manninen, Fröjd, Haravuori, Lindberg, Marttunen, Kettunen and Therman2016; Zijlema et al., Reference Zijlema, Stolk, Löwe, Rief, White and Rosmalen2013). The potential predictors for the persistence of common somatic symptoms, all assessed at baseline, are described in Table 1. Femininity was operationalized via a recently developed gender index (Ballering et al., Reference Ballering, Bonvanie, Olde Hartman, Monden and Rosmalen2020), which accounts for the time-, place-, and society-bound nature of gender. In a subsample of adult Lifelines participants, with no suspected intersex condition or non-conform gender identity, a LASSO logistic regression model that predicts the participants' sex by means of psychosocial characteristics, including but not limited to hobbies, type of profession, dietary preferences, and time spent on household tasks, was calculated. In total, 85 unique psychosocial variables were included in the model (AUC = 92%) and thus gendered. The included psychosocial predictors cover predominantly gender roles, and therefore reflect the gender roles as adhered to in the Lifelines cohort. The obtained estimates of the regression coefficients were applied to all adult Lifelines participants, providing each participant with an individual score on the gender index, i.e. participants' adherence to the gendered psychosocial variables. The gender index ranges from 0%, equaling masculinity, to 100%, equaling femininity. We assessed the multicollinearity of the predictor variables by means of the variance inflation factor (VIF). We found no indication of problems with multicollinearity, as VIF was <5 in all analyses (Miles & Shevlin, Reference Miles and Shevlin2001).

Table 1. Overview of potential predictors for persistent common somatic symptoms (all assessed at baseline)

Statistical analyses

To identify different developmental trajectories of common somatic symptoms over time, latent class trajectory modeling was conducted in R version 3.5.2 and R studio 1.1.383 (R package ‘lcmm’, version 1.7.8) (Proust-Lima, Philipps, & Liquet, Reference Proust-Lima, Philipps and Liquet2017). Notably, these trajectories should not be reified, as these are merely estimated latent groups, not actual observed groups. An advantage of latent class trajectory modeling using full information maximum likelihood estimation as applied in this study is that it allows the number of times a participant was assessed to vary between participants, which facilitates the inclusion of participants with intermittent missing data or those who dropped out (Proust-Lima et al., Reference Proust-Lima, Philipps and Liquet2017). We used the GRoLTS guidelines and Lennon's et al. framework as a guidance to construct and interpret latent class trajectory modeling (Lennon et al., Reference Lennon, Kelly, Sperrin, Buchan, Cross, Leitzmann and Renehan2018; Van De Schoot et al., Reference Van De Schoot, Sijbrandij, Winter, Depaoli and Vermunt2017).

Latent classes with different growth trajectories were modeled based on growth models that define how an outcome changes as a function of time, using an intercept and slope parameter. In order to find the model that best described the data, we fitted latent class growth models with fixed class-specific intercepts and slopes (LCGA), as well as more flexible growth mixture models (GMM) with (i) a random class-specific intercept and fixed slope per class and (ii) random class-specific intercepts and slopes. We fitted models with both linear and quadratic trajectories. LCGA and GMM models were fitted to the data with increasing number of classes (g = 1 to g = 7), after which indices of model fit were compared. Every model was run with multiple (25) random start values (derived from the one-class model) in order to identify a replicable Log-Likelihood maximum, that was unlikely to be at a local maximum. The best-fitting model was then fully fitted with a maximum of 500 iterations. In the models, the intercept and slope variances were constrained to be equal across classes. Data were rearranged as a function of chronological months since inclusion into the study. This resulted in 123 (baseline measurement being 0 months, the latest measurement being 123 months) instead of four assessment points, allowing for more complex trajectories to be modeled. Data points at which no valid information on the SCL-90 SOM was provided were excluded (N = 39 (0.02%), N = 767 (0.62%), N = 309 (0.33%), and N = 320 (0.36%) participants at the first, second, third, and fourth measurement, respectively). Ultimately, participants were allocated to a class based on their highest posterior class probability score. Participants with low posterior probabilities for all classes (<0.50) were excluded from the analyses.

The models with increasing number of classes were compared on four a priori formulated criteria: (i) the model with the lowest Bayesian Information Criterion (BIC) value was favorable (Nylund, Asparouhov, & Muthén, Reference Nylund, Asparouhov and Muthén2007; Van De Schoot et al., Reference Van De Schoot, Sijbrandij, Winter, Depaoli and Vermunt2017); (ii) the entropy of the model with the lowest BIC was assessed, as high entropy (>0.80) indicates strong distinctive capabilities between trajectory classes (Celeux & Soromenho, Reference Celeux and Soromenho1996); (iii) class sizes, as class sizes should not comprise less than 1% of the sample (Infurna & Grimm, Reference Infurna and Grimm2017); and (iv) theoretical plausibility, for example, verifying whether the observed trajectories fit the longitudinal plots of raw data and previous empirical findings (Ram & Grimm, Reference Ram and Grimm2009).

To identify the predictors of somatic symptom trajectories, we conducted multiple logistic regression analyses, including all predictors as mentioned in Table 1 as independent variables. To study whether the predictors for the latent subgroups differed between females and males, we included interaction terms between sex and predictors.

To test whether the continuous covariates included in the multiple logistic regression analyses fulfilled the linearity assumption of multiple logistic regression, we divided the covariates into quartiles, and assessed whether the estimates increased or decreased monotonically. IBM SPSS v. 25 was used to perform regression analyses. We maintained a two-sided α-value, corrected for multiple comparisons, of 0.001 (0.05/47, 24 predictors and 23 sex-by-predictor interaction terms within a family of tests).

Three sensitivity analyses were performed. First, we performed the regression analyses without adjusting for the presence of chronic diseases to explore its influence on the association between predictors and the identified trajectories. Second, we assessed whether the association between negative life events and the observed trajectories was partly explained by health-related negative life events. We excluded any health-related negative life events from the regression analyses to assess its influence on the association between negative life events and the identified trajectories. Third, we performed regression analyses with different symptom trajectories as a reference category.

Results

We found that the mean SCL-90 SOM score in the complete sample remained stable over time with scores of 1.36 (s.d. = 0.38), 1.38 (s.d. = 0.45), 1.42 (s.d. = 0.44), and 1.36 (s.d. = 0.42) at subsequent measurement waves.

Trajectory modeling

Of the fitted models, LCGA performed best (online Supplementary Appendix C). Therefore, only estimates of LCGA models are shown in Table 2. The five-class model fitted the data best, as is indicated by the lowest BIC value, good entropy, and acceptable class sizes. The class-specific predicted mean SCL-90 SOM trajectories are displayed in Fig. 1. The first class, which comprises the majority of the population (N = 113 444; 75.4%) reported minimal to no SCL-90 SOM symptoms over time. The second class (N = 1717; 1.1%) reported a high, stable SCL-90 SOM symptom score over time. The third class (N = 7168; 4.8%) showed slightly decreasing, intermediate SCL-90 SOM symptom score, whereas the fourth class (N = 25 954; 17.3%) showed a low, stable SCL-90 SOM symptom score, albeit somewhat higher than the first class. The fifth class (N = 2211; 1.5%) started with a relatively low SCL-90 SOM score, which steeply increased over time. Online Supplementary Appendix D shows plots with individual SCL-90 SOM score trajectories, stratified per class.

Fig. 1. Class-specific mean predicted symptom trajectories.

Table 2. Parameter estimates for 1–7 classes (N = 150 494) using a linear trajectory function

a Bayesian Information Criteria.

b Preferred model.

Logistic regression analyses

Participants were allocated to one of the trajectory classes, based on their posterior class probability score; 1495 (1.0%) participants with low posterior probabilities for all classes (<0.50) were excluded from the analyses.

First, we identified the predictors of high, stable symptom severity (class 2) compared with low, stable symptom severity (class 4) by multiple logistic regression analyses (Table 3). Class 4 was selected as the reference category to facilitate a comparison with the subsequent analyses. We found that female sex was significantly associated with high symptom severity (OR 1.74, 95% CI 1.36–2.22), while femininity was associated with low symptom severity (OR 0.60, 95% CI 0.44–0.82). Also, increased physical functioning and emotional wellbeing were associated with low symptom severity (OR 0.96, 95% CI 0.96–0.97 for both predictors). On the other hand, better self-rated health was associated with high symptom severity (OR 1.03, 95% CI 1.02–1.04). We found that personality traits were not statistically significantly associated with high symptom severity. We assessed the statistical significance of the interaction term between sex and all predictors. Only the interaction terms between sex and hours of paid employment (OR 0.98, 95% CI 0.98–0.99), and sex and physical functioning (OR 0.99, 95% CI 0.98–0.99) were statistically significant, indicating that the negative association between these predictors and a high symptom severity was stronger in females than in males.

Table 3. The associations between predictors and high, stable symptom severity over time

a Please note that the odds presented are per unit change on the scale of the predictor, thus magnitudes are not always directly comparable. *Indicates statistical significance (p < 0.001). Note: Nagelkerke's R 2 for the model including all participants, only the men, and only the women allocated to class 2 and class 4 are 0.38, 0.36, and 0.39, respectively.

b Interaction terms between these predictors and sex were statistically significant.

Second, we assessed which predictors were associated with increasing symptom severity, by multiple logistic regression analyses predicting increasing SCL-90 SOM score over time (class 5) versus low, stable SCL-90 SOM score (class 4), as these trajectories had a similar intercept. This allows for identification of predictors that may associate with an increasing symptom severity, instead of a low, stable symptom severity. Table 4 shows that females have 1.15 times the odds (95% CI 0.99–1.40) compared to males of having increasing symptom severity over time, however this result did not reach statistical significance. Femininity seemed to be protective of increasing symptom severity over time (OR 0.66, 95% CI 0.51–0.85), yet experiencing a negative life event is disadvantageous (OR 1.30, 95% CI 1.16–1.47). The OR of the interaction term between sex and education (OR 1.23, 95% CI 1.01–1.58), sex and the presence of chronic disease (OR 0.79, 95% CI 0.63–0.98), sex and physical functioning (OR 0.99, 95% CI 0.98–0.99), and sex and the score on the NEO-PI-R deliberation subscale (OR 1.35, 95% CI 1.10–1.64) were statistically significant, indicating that the association between these predictors and increasing symptom severity differed between females and males. The results of the sensitivity analyses, which assess the effect of the presence of chronic diseases and health-related negative life events on the association between femininity or sex and symptom trajectories, as well as the effect of selecting class 1 (no, stable symptoms) as the reference symptom trajectory instead of class 4 (low, stable symptoms) yielded essentially the same results and are shown in online Supplementary Appendix E.

Table 4. The associations between multiple predictors and increasing symptom severity over time

a Please note that the odds presented are per unit change on the scale of the predictor, thus magnitudes are not always directly comparable. *Indicates statistical significance (p < 0.001). Nagelkerke's R 2 for the model including all participants, only the men, and only the women allocated to class 5 and class 4 are 0.11, 0.28, and 0.11, respectively.

b Interaction terms between these predictors and sex were statistically significant.

Discussion

To the best of our knowledge, this is the first large general population cohort study that assesses longitudinal somatic symptom trajectories by means of LCGA. This data-driven method allows for the identification of homogeneous patterns of symptom severity over time from heterogeneous data. We found that a five-class linear model that excluded intraclass individual variation fitted the data best. The majority of the cohort had a stable symptom trajectory (93.7%), with low (class 1; 75.4%), slightly higher (class 4; 17.3%), and high (class 2; 1.1%) symptom severity. In addition, we identified a class with slightly decreasing (class 3; 4.8%) and a class with increasing (class 5; 1.5%) symptom severity over time. We found that female sex is a predictor for a high, stable SCL-90 SOM score. However, female sex only approached statistical significance for an increasing SCL-90 SOM score, compared to male sex. Femininity, in contrast, appeared to be protective for both a stable and an increasing somatic symptom severity. In females, hours of paid employment and physical functioning were more strongly negatively associated with stable symptom severity than in males. Regarding increasing symptom severity over time, education, the presence of chronic disease, physical functioning and the score on the NEO-PI-R deliberation subscale differed in predictive strength between females and males.

Strengths and limitations

The principal strength of this study is the data-driven approach we used to estimate common somatic symptom trajectories in a large general population cohort. Latent class trajectory modeling allows for identifying nuances between seemingly similar subpopulations (Lennon et al., Reference Lennon, Kelly, Sperrin, Buchan, Cross, Leitzmann and Renehan2018). This inductive approach facilitates the identification of novel predictors of at-risk subpopulations, especially if individuals' symptom trajectories are analyzed as an outcome, rather than as an independent variable. However, note that when analysis involves latent class trajectories either as outcome or exposure, one should not view the trajectories as concrete entities, but rather as a method to reduce the observed heterogeneity in the data (van Loo, Wanders, Wardenaar, & Fried, Reference van Loo, Wanders, Wardenaar and Fried2018). Furthermore, our cohort had a large sample size and was followed up for a long period of time, with multiple measurements, allowing for complex models to be fitted. Lastly, the incorporation of participants' gender is advantageous, as it allows for disentangling the biological and psychosocial influences related to being a woman or man on somatic symptom trajectories.

Our study had several limitations. First, we assessed the mean SCL-90 SOM score as an aggregate score and therefore we have not differentiated between individual symptoms. Possibly, participants had different symptoms that bothered them over time, despite their symptom scores remaining stable. Second, we could not account for illness cognitions or health care utilization, as no data hereon were available. Illness cognitions are thought to account for 30–40% of the variance in health outcomes related to somatic symptoms (McAndrew et al., Reference McAndrew, Crede, Maestro, Slotkin, Kimber and Phillips2019), such as the persistence of symptoms (Moss-Morris, Spence, & Hou, Reference Moss-Morris, Spence and Hou2011). Similarly, health care utilization is known to associate somatic symptom burden, and thus may affect symptom trajectories (Lee, Creed, Ma, & Leung, Reference Lee, Creed, Ma and Leung2015). Also, we only assessed predictors at baseline, but predictors may also have an influence during the course of one's symptoms, such as the development of chronic diseases or health care utilization. All predictors are self-reported, which means that the measures of the life-time prevalence of mood and anxiety disorders are not necessarily a clinical diagnosis, and the latter two may be more reflective of experienced mood and anxiety symptoms than of a clinical diagnosis.

Latent class trajectories

We identified three stable symptom severity trajectories (93.7%) and two relatively small classes that follow a consistently increasing (1.5%) and decreasing (4.8%) course. The proportion of participants with non-stable trajectories is in line with previous research that used latent class trajectory modeling. In a patient population with medically unexplained somatic symptoms, 92.6% of the patients had a stable symptom score over the 2 years follow-up. The remaining 7.4% of patients improved. However, this study used the Patient Health Questionnaire-15 to measure symptom severity, which differs from the current SCL-90 SOM subscale (Claassen-van Dessel et al., Reference Claassen-van Dessel, van der Wouden, Hoekstra, Dekker and van der Horst2018). Another study conducted in a general adolescent cohort showed that 85.3% of adolescents had a predominantly stable symptom trajectory over time (Janssens et al., Reference Janssens, Klis, Kingma, Oldehinkel and Rosmalen2014). Four trajectories were identified in this study. Again, this study used a different questionnaire, but more importantly, an adolescent cohort might not be directly comparable to an adult cohort. It is disputed whether somatic symptoms in adolescents and adults are comparable in onset, healthcare-seeking behavior, and treatment, possibly due to stronger family influences in adolescents and a differing physiology compared to adults (Weisblatt, Hindley, & Rask, Reference Weisblatt, Hindley, Rask, Creed, Henningsen and Fink2011). Furthermore, adolescence is thought to be accompanied by a heightened bodily awareness and therefore with the experience of common somatic symptoms (Rhee, Reference Rhee2005). Overall, stable somatic symptom scores over time prevail in the aforementioned studies despite the differing study populations.

Sex and gender in relation to somatic symptom trajectories

In line with our current study, recent studies have found female sex to be associated with more numerous and more severe somatic symptoms (Ballering et al., Reference Ballering, Bonvanie, Olde Hartman, Monden and Rosmalen2020; Tomenson et al., Reference Tomenson, Essau, Jacobi, Ladwig, Leiknes, Lieb and Rief2013), as well as with an increasing severity of somatic symptoms over time (De Gucht et al., Reference De Gucht, Fischler and Heiser2004; Janssens et al., Reference Janssens, Klis, Kingma, Oldehinkel and Rosmalen2014). Multiple explanations have been raised for this phenomenon. First, females may have a heightened pain sensitivity due to biological differences (Fillingim, Reference Fillingim and Legato2017). Sex hormones, genotypes, immune systems, and neurology may induce differences in the processing of pain that predispose females to worse symptom trajectories than males (Bartley & Fillingim, Reference Bartley and Fillingim2013). Second, females are thought to be more aware of bodily sensations than males. This heightened awareness allows for easier and earlier perception of somatic symptoms in females than in males (Barsky, Peekna, & Borus, Reference Barsky, Peekna and Borus2001). These biological differences may explain the female preponderance in somatic symptoms as found in our study, but our results also point toward a role for psychosocial gender differences.

Femininity was found to be protective against both a high and increasing symptom severity. This is different from earlier cross-sectional studies that showed an association between femininity, measured by the gender index, domestic responsibilities, or the BEM sex role inventory, respectively, and higher levels of common somatic symptoms (Ballering et al., Reference Ballering, Bonvanie, Olde Hartman, Monden and Rosmalen2020; Krantz & Ostergren, Reference Krantz and Ostergren2001) or that found no association (Castro, Carbonell, & Anestis, Reference Castro, Carbonell and Anestis2012). These differences may be explained by the longitudinal nature of the current study, which provides insight into the dynamics of symptoms over time and may result in a more precise assessment of somatic symptom severity. Furthermore, in the former study in which the gender index was used, all adult Lifelines participants were included and femininity was found to be associated with more severe symptoms. However, the current study focusses on participants with high or increasing symptom severity (1.1% and 1.5% of the adult participants, respectively) and femininity was not found to associate with increased symptom severity. Possibly increased healthcare-seeking behavior plays a role in femininity being a protective factor for both high, stable and increasing symptom severity over time (Steinbrecher & Hiller, Reference Steinbrecher and Hiller2011), as healthcare-seeking behavior is known to be gendered (Hart, Saperstein, Magliozzi, & Westbrook, Reference Hart, Saperstein, Magliozzi and Westbrook2019) and may prevent worsening of symptoms over time (Lee et al., Reference Lee, Creed, Ma and Leung2015). Feminine people are thought to have a lower threshold to seek help or medical care, especially from their GP (Loikas et al., Reference Loikas, Karlsson, von Euler, Hallgren, Schenck-Gustafsson and Bastholm Rahmner2015). Femininity is, for example, related to providing and facilitating care for the family, allowing feminine people to be more often in contact with healthcare providers, concomitantly lowering the barrier for healthcare-seeking behavior. Additionally, femininity is related to being open and less stoic about one's symptoms, facilitating healthcare-seeking behavior. Masculinity, in contrast, relates to being less expressive about distress and seeking help for symptoms is stereotypically seen as socially undermining an individual's masculinity (Barsky et al., Reference Barsky, Peekna and Borus2001; MacLean, Sweeting, & Hunt, Reference MacLean, Sweeting and Hunt2010). The earlier study that used the gender index included participants with low symptom severity as well, in which healthcare-seeking behavior may not be as important.

In addition, what constitutes femininity differs between studies and changes over time and place, yielding different results. For example, the BEM sex role inventory was developed in 1974 and then widely applied, but is currently deemed to hold limited validity as an operationalization of femininity or masculinity (Donnelly & Twenge, Reference Donnelly and Twenge2017). Lastly, the association between femininity and higher levels of somatic symptom severity in the previous studies may have been partially explained by the presence of chronic diseases, whilst we adjusted for this in the current study. Sensitivity analyses showed that indeed an adjustment for the presence of chronic diseases slightly strengthened the protective association between femininity and high or increasing symptom severity over time.

Predictors of stable and increasing common somatic symptoms

In addition to sex and gender, multiple factors were predictive of persistent common somatic symptoms. Higher education, higher levels of physical functioning, and higher emotional wellbeing at baseline are associated with low, stable symptom severity. This is in line with previous research that suggests that these factors have a positive influence on overall functioning and wellbeing (van Eck van der Sluijs et al., Reference van Eck van der Sluijs, ten Have, de Graaf, Rijnders, van Marwijk and van der Feltz-Cornelis2018). Here, we also found that the lower one rates his or her own general health, the lower the odds that one has persisting common somatic symptoms. Perceptions of low general health may prompt one to seek medical help, which may lead to an improvement of symptoms (Steinbrecher & Hiller, Reference Steinbrecher and Hiller2011). However, previous studies contradict each other with regards to self-rated health and the course of somatic symptoms (Janssens et al., Reference Janssens, Klis, Kingma, Oldehinkel and Rosmalen2014; Steinbrecher & Hiller, Reference Steinbrecher and Hiller2011; van Eck van der Sluijs et al., Reference van Eck van der Sluijs, ten Have, de Graaf, Rijnders, van Marwijk and van der Feltz-Cornelis2018). These contradictions might be due to the different conceptualizations of self-related health and somatic symptoms in the studies.

We also found that the presence of anxiety disorders is related to stable and increasing symptom severity in females. Anxiety disorders are diagnosed approximately twice as often in females than in males (McLean, Asnaani, Litz, & Hofmann, Reference McLean, Asnaani, Litz and Hofmann2011), and often manifest themselves with prominent somatic characteristics (Bekhuis, Schoevers, Van Borkulo, Rosmalen, & Boschloo, Reference Bekhuis, Schoevers, Van Borkulo, Rosmalen and Boschloo2016). Therefore, the presence of anxiety disorders may contribute to the elevated SCL-90 SOM scores as found in this study. A self-reported mood disorder, however, was not associated with a high or increasing symptom severity. Evidence from cross-sectional studies suggests that mood disorders are associated with more severe common somatic symptoms (Bekhuis, Boschloo, Rosmalen, & Schoevers, Reference Bekhuis, Boschloo, Rosmalen and Schoevers2015; Löwe et al., Reference Löwe, Spitzer, Williams, Mussell, Schellberg and Kroenke2008). In contrast, results of longitudinal research assessing mood disorders in relation to somatic symptoms are contradictive (Niles & O'Donovan, Reference Niles and O'Donovan2019; Steinbrecher & Hiller, Reference Steinbrecher and Hiller2011; van Eck van der Sluijs et al., Reference van Eck van der Sluijs, ten Have, de Graaf, Rijnders, van Marwijk and van der Feltz-Cornelis2018). To date, it has not been possible to draw any definitive conclusion on whether mood disorders predict an unfavorable somatic symptom prognosis. It has been argued that a similar mechanism as mentioned above may apply to people with mood disorders: being affected by mood disorders may prompt healthcare-seeking behavior, resulting in an improvement of symptoms. The aforementioned longitudinal studies, however, including our study, do not differentiate between, or assess different, somatic symptoms. Thus the association between mood disorders and one type of symptom may be overshadowed by the lack of an association with other types of symptoms.

We also found that negative life events are predictors of increasing symptom severity. As a sensitivity analysis, we removed any item from the negative life events scale that was related to experiencing a severe disease . The direction and strength of the association remained similar. It is thought that psychological distress as a consequence of a negative life event may result in somatic symptoms (Tak, Kingma, van Ockenburg, Ormel, & Rosmalen, Reference Tak, Kingma, van Ockenburg, Ormel and Rosmalen2015). Physiological and emotional stress mechanisms are suggested as the link between psychological distress and somatic symptoms (Bonvanie, Janssens, Rosmalen, & Oldehinkel, Reference Bonvanie, Janssens, Rosmalen and Oldehinkel2017). Such mechanisms heighten one's bodily vigilance, consequently facilitating people to interpret bodily signals more easily as somatic symptoms.

Implications for further research and clinical practice

Further research could focus on symptom-specific latent class trajectories, to assess whether differences in trajectories exist between symptoms. Additionally, one could study whether the type of reported symptoms change over time in the classes with stable, high mean symptom severity scores and whether these changes follow specific sequences. The results from this study also show that the majority of the general population remains stable in their level of symptom severity and that only a relatively small proportion has a high or increasing symptom severity. However, it remains unknown whether it is merely the latter population that seeks medical attention and if so, what factors are associated with this healthcare-seeking behavior. As a protective association between femininity and a high and increasing symptom severity was found in this study, it is especially interesting to study to what extent femininity relates to healthcare-seeking behavior for somatic symptoms. For those patients who visit their GP, it is pivotal that predictors for increasing symptom severity are recognized, preferably in an early stage.

We found no large sex differences in the predictors of high or increasing symptom severity, thus it may not be clinically useful to distinguish between predictors specific to male or female patients withpersistent common somatic symptoms. Furthermore, for reasons of clarity, we currently described the associations of sex and gender with common somatic symptom trajectories separately. However, although sex and gender are different concepts, a clear demarcation between these in clinical practice is artificial: clinicians cannot consider sex without gender and vice versa as these concepts are intertwined.

Supplementary material

The supplementary material for this article can be found at https://doi.org/10.1017/S0033291720004043.

Acknowledgements

The Lifelines Biobank initiative has been made possible by subsidy from the Dutch Ministry of Health, Welfare and Sport, the Dutch Ministry of Economic Affairs, the University Medical Center Groningen (UMCG the Netherlands), University Groningen and the Northern Provinces of the Netherlands. The authors wish to acknowledge the services of the Lifelines Cohort Study, the contributing research centers delivering data to Lifelines, and all the study participants.

Financial support

This work was supported by ZonMw (project number 849200013).

Conflict of interest

None.

References

Aaronson, N. K., Muller, M., Cohen, P. D. A., Essink-Bot, M., Fekkes, M., Sanderman, R., … Verrips, E. (1998). Translation, validation, and norming of the Dutch language version of the SF-36 health survey in community and chronic disease populations. Journal of Clinical Epidemiology, 51(11), 10551068. doi:10.1016/S0895-4356(98)00097-3.CrossRefGoogle ScholarPubMed
Ballering, A. V., Bonvanie, I. J., Olde Hartman, T. C., Monden, R., & Rosmalen, J. G. M. (2020). Gender and sex independently associate with common somatic symptoms and lifetime prevalence of chronic disease. Social Science & Medicine, 253, 112968. doi:10.1016/j.socscimed.2020.112968.CrossRefGoogle ScholarPubMed
Barsky, A. J., Peekna, H. M., & Borus, J. F. (2001). Somatic symptom reporting in women and men. Journal of General Internal Medicine, 16(4), 266275.CrossRefGoogle ScholarPubMed
Bartley, E. J., & Fillingim, R. B. (2013). Sex differences in pain: A brief review of clinical and experimental findings. British Journal of Anaesthesia, 111(1), 5258.CrossRefGoogle ScholarPubMed
Bekhuis, E., Boschloo, L., Rosmalen, J. G. M., & Schoevers, R. A. (2015). Differential associations of specific depressive and anxiety disorders with somatic symptoms. Journal of Psychosomatic Research, 78(2), 116122. doi:10.1016/j.jpsychores.2014.11.007.CrossRefGoogle ScholarPubMed
Bekhuis, E., Schoevers, R., Van Borkulo, C., Rosmalen, J., & Boschloo, L. (2016). The network structure of major depressive disorder, generalized anxiety disorder and somatic symptomatology. Psychological Medicine, 46(14), 29892998. doi:10.1017/S0033291716001550.CrossRefGoogle Scholar
Bonvanie, I., Janssens, K., Rosmalen, J., & Oldehinkel, A. (2017). Life events and functional somatic symptoms: A population study in older adolescents. British Journal of Psychology, 108(2), 318333. doi:10.1111/bjop.12198.CrossRefGoogle ScholarPubMed
Bonvanie, I. J., Rosmalen, J. G. M., van Rhede van der Kloot, C. M., Oldehinkel, A. J., & Janssens, K. A. M. (2015). Functional somatic symptoms are associated with perfectionism in adolescents. Journal of Psychosomatic Research, 79(4), 328330. doi:10.1016/j.jpsychores.2015.07.009.CrossRefGoogle ScholarPubMed
Brugha, T. S., & Cragg, D. (1990). The list of threatening experiences: The reliability and validity of a brief life events questionnaire. Acta Psychiatrica Scandinavica, 82(1), 7781.CrossRefGoogle ScholarPubMed
Budtz-Lilly, A., Vestergaard, M., Fink, P., Carlsen, A. H., & Rosendal, M. (2015). The prognosis of bodily distress syndrome: A cohort study in primary care. General Hospital Psychiatry, 37(6), 560566. doi:10.1016/j.genhosppsych.2015.08.002CrossRefGoogle ScholarPubMed
Castro, Y., Carbonell, J. L., & Anestis, J. C. (2012). The influence of gender role on the prediction of antisocial behaviour and somatization. The International Journal of Social Psychiatry, 58(4), 409416. doi:10.1177/0020764011406807.CrossRefGoogle ScholarPubMed
Celeux, G., & Soromenho, G. (1996). An entropy criterion for assessing the number of clusters in a mixture model. Journal of Classification, 13(2), 195212.CrossRefGoogle Scholar
Claassen-van Dessel, N., van der Wouden, J. C., Hoekstra, T., Dekker, J., & van der Horst, H. E. (2018). The 2-year course of medically unexplained physical symptoms (MUPS) in terms of symptom severity and functional status: Results of the PROSPECTS cohort study. Journal of Psychosomatic Research, 104, 7687. 10.1016/j.jpsychores.2017.11.012..CrossRefGoogle ScholarPubMed
Costa, P., & McCrae, R. (1992). Revised NEO personality inventory (NEO-PI-R) and NEO five-factor inventory (NEO-FFI) professional manual. Odessa, FL: Psychological assessment resources.Google Scholar
De Gucht, V., Fischler, B. F., & Heiser, W. (2004). Personality and affect as determinants of medically unexplained symptoms in primary care; a follow-up study. Journal of Psychosomatic Research JID - 0376333, 56(3), 279285.CrossRefGoogle ScholarPubMed
Dirkzwager, A. J., & Verhaak, P. F. (2007). Patients with persistent medically unexplained symptoms in general practice: Characteristics and quality of care. BMC Family Practice, 8(1), 33.CrossRefGoogle ScholarPubMed
Donnelly, K., & Twenge, J. M. (2017). Masculine and feminine traits on the BEM sex-role inventory, 1993–2012: A cross-temporal meta-analysis. Sex Roles, 76(9), 556565. doi:10.1007/s11199-016-0625-y.CrossRefGoogle Scholar
Dutch Ministry of Health, Welfare and Sports (2019). Ranglijst aandoeningen op basis van verlies aan gezonde levensjaren (ziektejaarequivalenten). Retrieved from https://www.volksgezondheidenzorg.info/ranglijst/ranglijst-aandoeningen-op-basis-van-verlies-aan-gezonde-levensjaren-ziektejaarequivalenten.Google Scholar
Fillingim, R. B. (2017). Sex, gender, and pain. In Legato, M. J. (Ed.), Principles of gender-specific medicine (3rd ed., pp. 481496). San Diego: Academic Press. doi:10.1016/B978-0-12-803506-1.00038-3.CrossRefGoogle Scholar
Gijsen, R., Poos, M. J. J. C., Slobbe, L. C., Mulder, M., In 't Panhuis-Plasmans, M. H. D., & Hoeymans, N. (2013). Een nieuwe selectie van ziekten voor de volksgezondheid toekomst verkenningen. (No. 010003004/2013). Bilthoven: RIVM.Google Scholar
Hart, C. G., Saperstein, A., Magliozzi, D., & Westbrook, L. (2019). Gender and health: Beyond binary categorical measurement. Journal of Health and Social Behavior, 60(1), 101118. doi:10.1177/0022146519825749.CrossRefGoogle ScholarPubMed
Heidari, S., Babor, T. F., De Castro, P., Tort, S., & Curno, M. (2016). Sex and gender equity in research: Rationale for the SAGER guidelines and recommended use. Research Integrity and Peer Review, 1(1), 2. doi:10.1186/s41073-016-0007-6.CrossRefGoogle ScholarPubMed
Hill, R. D., Van Boxtel, M., Ponds, R., Houx, P., & Jolles, J. (2005). Positive affect and its relationship to free recall memory performance in a sample of older Dutch adults from the Maastricht aging study. International Journal of Geriatric Psychiatry, 20(5), 429435.CrossRefGoogle Scholar
Hoeymans, N., Gijsen, R., & Slobbe, L. C. (2013). 59 Important health problems; a selection of diseases for public health monitoring. Nederlands Tijdschrift Voor Geneeskunde, 157(31), A5994.Google ScholarPubMed
Infurna, F. J., & Grimm, K. J. (2017). The use of growth mixture modeling for studying resilience to major life stressors in adulthood and old age: Lessons for class size and identification and model selection. The Journals of Gerontology: Series B, 73(1), 148159.CrossRefGoogle ScholarPubMed
Janssens, K. A. M., Klis, S., Kingma, E. M., Oldehinkel, A. J., & Rosmalen, J. G. M. (2014). Predictors for persistence of functional somatic symptoms in adolescents. The Journal of Pediatrics, 164(4), 900905, e2.CrossRefGoogle ScholarPubMed
Johnson, J. L., Greaves, L., & Repta, R. (2009). Better science with sex and gender: Facilitating the use of a sex and gender-based analysis in health research. International Journal for Equity in Health, 8(1), 14.CrossRefGoogle ScholarPubMed
Joustra, M. L., Janssens, K. A. M., Bültmann, U., & Rosmalen, J. G. M. (2015). Functional limitations in functional somatic syndromes and well-defined medical diseases. Results from the general population cohort LifeLines. Journal of Psychosomatic Medicine, 79(2), 9499. doi:10.1016/j.jpsychores.2015.05.004.CrossRefGoogle ScholarPubMed
Klijs, B., Scholtens, S., Mandemakers, J. J., Snieder, H., Stolk, R. P., & Smidt, N. (2015). Representativeness of the LifeLines cohort study. PLoS ONE, 10(9), e0137203.CrossRefGoogle ScholarPubMed
Koch, H., van Bokhoven, M., Bindels, P., van der Weijden, T., Dinant, G. J., & ter Riet, G. (2009). The course of newly presented unexplained complaints in general practice patients: A prospective cohort study. Family Practice, 26(6), 455465. doi:10.1093/fampra/cmp067.CrossRefGoogle ScholarPubMed
Konnopka, A., Schaefert, R., Heinrich, S., Kaufmann, C., Luppa, M., Herzog, W., & König, H. (2012). Economics of medically unexplained symptoms: A systematic review of the literature. Psychotherapy and Psychosomatics, 81(5), 265275. doi:10.1159/000337349.CrossRefGoogle ScholarPubMed
Kooiman, C. G., Bolk, J. H., Rooijmans, H. G. M., & Trijsburg, R. W. (2004). Alexithymia does not predict the persistence of medically unexplained physical symptoms. Psychosomatic Medicine, 66(2), 224232.CrossRefGoogle Scholar
Krantz, G., & Ostergren, P. (2001). Double exposure: The combined impact of domestic responsibilities and job strain on common symptoms in employed Swedish women. The European Journal of Public Health, 11(4), 413419.CrossRefGoogle ScholarPubMed
Lamahewa, K., Buszewicz, M., Walters, K., Marston, L., & Nazareth, I. (2019). Persistent unexplained physical symptoms: A prospective longitudinal cohort study in UK primary care. The British Journal of General Practice, 69(681), e246e253. doi:10.3399/bjgp19X701249.CrossRefGoogle ScholarPubMed
Lee, S., Creed, F. H., Ma, Y., & Leung, C. M. (2015). Somatic symptom burden and health anxiety in the population and their correlates. Journal of Psychosomatic Research, 78(1), 7176. doi:10.1016/j.jpsychores.2014.11.012.CrossRefGoogle ScholarPubMed
Lennon, H., Kelly, S., Sperrin, M., Buchan, I., Cross, A. J., Leitzmann, M., … Renehan, A. G. (2018). Framework to construct and interpret latent class trajectory modelling. BMJ Open, 8(7), e020683. doi:10.1136/bmjopen-2017-020683.CrossRefGoogle ScholarPubMed
Loikas, D., Karlsson, L., von Euler, M., Hallgren, K., Schenck-Gustafsson, K., & Bastholm Rahmner, P. (2015). Does patient's sex influence treatment in primary care? Experiences and expressed knowledge among physicians–a qualitative study. BMC Family Practice, 16, 137. doi:10.1186/s12875-015-0351-5.CrossRefGoogle ScholarPubMed
Löwe, B., Spitzer, R. L., Williams, J. B., Mussell, M., Schellberg, D., & Kroenke, K. (2008). Depression, anxiety and somatization in primary care: Syndrome overlap and functional impairment. General Hospital Psychiatry, 30(3), 191199. doi:10.1016/j.genhosppsych.2008.01.001.CrossRefGoogle ScholarPubMed
MacLean, A., Sweeting, H., & Hunt, K. (2010). ‘Rules’ for boys, ‘guidelines’ for girls: Gender differences in symptom reporting during childhood and adolescence. Social Science & Medicine, 70(4), 597604.CrossRefGoogle ScholarPubMed
McAndrew, L. M., Crede, M., Maestro, K., Slotkin, S., Kimber, J., & Phillips, L. A. (2019). Using the common-sense model to understand health outcomes for medically unexplained symptoms: A meta-analysis. Health Psychology Review, 13(4), 427446.CrossRefGoogle ScholarPubMed
McLean, C. P., Asnaani, A., Litz, B. T., & Hofmann, S. G. (2011). Gender differences in anxiety disorders: Prevalence, course of illness, comorbidity and burden of illness. Journal of Psychiatric Research, 45(8), 10271035.CrossRefGoogle ScholarPubMed
Miles, J., & Shevlin, M. (2001). Applying regression and correlation: A guide for students and researchers (pp. 126135) London, England: Sage.Google Scholar
Moss-Morris, R., Spence, M., & Hou, R. (2011). The pathway from glandular fever to chronic fatigue syndrome: Can the cognitive behavioural model provide the map? Psychological Medicine, 41(5), 10991107.CrossRefGoogle ScholarPubMed
Niles, A. N., & O'Donovan, A. (2019). Comparing anxiety and depression to obesity and smoking as predictors of major medical illnesses and somatic symptoms. Health Psychology, 38(2), 172181. doi:10.1037/hea0000707.CrossRefGoogle ScholarPubMed
Nylund, K. L., Asparouhov, T., & Muthén, B. O. (2007). Deciding on the number of classes in latent class analysis and growth mixture modeling: A Monte Carlo simulation study. Structural Equation Modeling, 14(4), 535569.CrossRefGoogle Scholar
Olde Hartman, T. C., Blankenstein, A. H., Molenaar, A. O., Bentz van den Berg, D., Van der Horst, H. E., Arnold, I. A., … Woutersen-Koch, H. (2013). NHG-standaard somatisch onvoldoende verklaarde lichamelijke klachten (SOLK). Huisarts En Wetenschap, 56(5), 222230.Google Scholar
Olde Hartman, T. C., Borghuis, M. S., Lucassen, P. L., van de Laar, F. A., Speckens, A. E., & van Weel, C. (2009). Medically unexplained symptoms, somatisation disorder and hypochondriasis: Course and prognosis. A systematic review. Journal of Psychosomatic Research, 66(5), 363377.CrossRefGoogle ScholarPubMed
Proust-Lima, C., Philipps, V., & Liquet, B. (2017). Estimation of extended mixed models using latent classes and latent processes: The R package lcmm. Journal of Statistical Software, 78(2). doi:10.18637/jss.v078.i02.CrossRefGoogle Scholar
Ram, N., & Grimm, K. J. (2009). Methods and measures: Growth mixture modeling: A method for identifying differences in longitudinal change among unobserved groups. International Journal of Behavioral Development, 33(6), 565576.CrossRefGoogle Scholar
Rhee, H. (2005). Relationships between physical symptoms and pubertal development. Journal of Pediatric Health Care, 19(2), 95103.Google ScholarPubMed
Rosmalen, J. G. M., Bos, E. H., & de Jonge, P. (2012). Validation of the long-term difficulties inventory (LDI) and the list of threatening experiences (LTE) as measures of stress in epidemiological population-based cohort studies. Psychological Medicine, 42(12), 25992608.CrossRefGoogle ScholarPubMed
Rytilä-Manninen, M., Fröjd, S., Haravuori, H., Lindberg, N., Marttunen, M., Kettunen, K., … Therman, S. (2016). Psychometric properties of the symptom checklist-90 in adolescent psychiatric inpatients and age- and gender-matched community youth. Child and Adolescent Psychiatry and Mental Health, 10(23). doi:10.1186/s13034-016-0111-x.CrossRefGoogle ScholarPubMed
Scholtens, S., Smidt, N., Swertz, M. A., Bakker, S. J., Dotinga, A., Vonk, J. M., … Stolk, R. P. (2015). Cohort profile: LifeLines, a three-generation cohort study and biobank. International Journal of Epidemiology, 44(4), 11721180.CrossRefGoogle ScholarPubMed
Steinbrecher, N., & Hiller, W. (2011). Course and prediction of somatoform disorder and medically unexplained symptoms in primary care. General Hospital Psychiatry, 33(4). doi:10.1016/j.genhosppsych.2011.05.002.CrossRefGoogle ScholarPubMed
Tak, L. M., Kingma, E. M., van Ockenburg, S. L., Ormel, J., & Rosmalen, J. G. M. (2015). Age- and sex-specific associations between adverse life events and functional bodily symptoms in the general population. Journal of Psychosomatic Research, 79(2), 112116. doi:10.1016/j.jpsychores.2015.05.013.CrossRefGoogle ScholarPubMed
Tomenson, B., Essau, C., Jacobi, F., Ladwig, K. H., Leiknes, K. A., Lieb, R., … Rief, W. (2013). Total somatic symptom score as a predictor of health outcome in somatic symptom disorders. The British Journal of Psychiatry, 203(5), 373380.CrossRefGoogle ScholarPubMed
Van De Schoot, R., Sijbrandij, M., Winter, S. D., Depaoli, S., & Vermunt, J. K. (2017). The GRoLTS-checklist: Guidelines for reporting on latent trajectory studies. Structural Equation Modeling: A Multidisciplinary Journal, 24(3), 451467.CrossRefGoogle Scholar
van Eck van der Sluijs, J. F., ten Have, M., de Graaf, R., Rijnders, C. A. T., van Marwijk, H. W. J., & van der Feltz-Cornelis, C. M. (2018). Predictors of persistent medically unexplained physical symptoms: Findings from a general population study. Frontiers in Psychiatry, 9, 613. doi:10.3389/fpsyt.2018.00613.CrossRefGoogle ScholarPubMed
van Loo, H. M., Wanders, R. B., Wardenaar, K. J., & Fried, E. I. (2018). Problems with latent class analysis to detect data-driven subtypes of depression. Molecular Psychiatry, 23(3), 495496.CrossRefGoogle ScholarPubMed
van Zon, S. K. R., Snieder, H., Bültmann, U., & Reijneveld, S. A. (2017). The interaction of socioeconomic position and type 2 diabetes mellitus family history: A cross-sectional analysis of the lifelines cohort and biobank study. BMJ Open, 7(4), e015275. doi:10.1136/bmjopen-2016-015275.CrossRefGoogle ScholarPubMed
Verhaak, P. F., Meijer, S. A., Visser, A. P., & Wolters, G. (2006). Persistent presentation of medically unexplained symptoms in general practice. Family Practice, 23(4), 414420.CrossRefGoogle ScholarPubMed
von Elm, E., Altman, D. G., Egger, M., Pocock, S. J., Gøtzsche, P. C., & Vandenbroucke, J. P. (2014). The strengthening the reporting of observational studies in epidemiology (STROBE) statement: Guidelines for reporting observational studies. International Journal of Surgery, 12(12). doi:10.1016/j.ijsu.2014.07.013.CrossRefGoogle ScholarPubMed
Weisblatt, E., Hindley, P., & Rask, C. U. (2011). Medically unexplained symptoms in children and adolescents. In Creed, F., Henningsen, P. & Fink, P. (Eds.), Medically unexplained symptoms, somatisation and bodily distress - developing better clinical services (1st ed., pp. 158174). Cambridge, UK: Cambridge University Press.CrossRefGoogle Scholar
Zijlema, W. L., Stolk, R. P., Löwe, B., Rief, W., White, P. D., & Rosmalen, J. G. M. (2013). How to assess common somatic symptoms in large-scale studies: A systematic review of questionnaires. Journal of Psychosomatic Research, 74(6), 459468.CrossRefGoogle ScholarPubMed
Figure 0

Table 1. Overview of potential predictors for persistent common somatic symptoms (all assessed at baseline)

Figure 1

Fig. 1. Class-specific mean predicted symptom trajectories.

Figure 2

Table 2. Parameter estimates for 1–7 classes (N = 150 494) using a linear trajectory function

Figure 3

Table 3. The associations between predictors and high, stable symptom severity over time

Figure 4

Table 4. The associations between multiple predictors and increasing symptom severity over time

Supplementary material: File

Ballering et al. Supplementary Materials

Ballering et al. Supplementary Materials

Download Ballering et al. Supplementary Materials(File)
File 279 KB
You have Access
Open access
3
Cited by

Save article to Kindle

To save this article to your Kindle, first ensure coreplatform@cambridge.org is added to your Approved Personal Document E-mail List under your Personal Document Settings on the Manage Your Content and Devices page of your Amazon account. Then enter the ‘name’ part of your Kindle email address below. Find out more about saving to your Kindle.

Note you can select to save to either the @free.kindle.com or @kindle.com variations. ‘@free.kindle.com’ emails are free but can only be saved to your device when it is connected to wi-fi. ‘@kindle.com’ emails can be delivered even when you are not connected to wi-fi, but note that service fees apply.

Find out more about the Kindle Personal Document Service.

Female sex and femininity independently associate with common somatic symptom trajectories
Available formats
×

Save article to Dropbox

To save this article to your Dropbox account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you used this feature, you will be asked to authorise Cambridge Core to connect with your Dropbox account. Find out more about saving content to Dropbox.

Female sex and femininity independently associate with common somatic symptom trajectories
Available formats
×

Save article to Google Drive

To save this article to your Google Drive account, please select one or more formats and confirm that you agree to abide by our usage policies. If this is the first time you used this feature, you will be asked to authorise Cambridge Core to connect with your Google Drive account. Find out more about saving content to Google Drive.

Female sex and femininity independently associate with common somatic symptom trajectories
Available formats
×
×

Reply to: Submit a response

Please enter your response.

Your details

Please enter a valid email address.

Conflicting interests

Do you have any conflicting interests? *