Authors: Sarah E Weber, Dóra Körmendiné Farkas, Sharon M Casey, Tina Bech Olesen, Mette Nørgaard, Jaimie L Gradus
Categories: Original Research, trauma and stressor related disorders, mental health, human papillomavirus, cancer, epidemiology, cohort studies
Source: Clinical Epidemiology
Doi: 10.2147/CLEP.S596452
Authors: Sarah E Weber, Dóra Körmendiné Farkas, Sharon M Casey, Tina Bech Olesen, Mette Nørgaard, Jaimie L Gradus
Stress disorders are associated with immunosuppression, potentially reducing the ability to clear human papillomavirus (HPV) infections, leading to an increased risk of HPV-related cancers. We investigated the association between stress disorders and HPV-related cancers in Denmark.
Using population-based medical registries, we identified patients with an incident hospital-diagnosed stress disorder between 1 January 1995 and 31 December 2021 (acute stress reaction, posttraumatic stress disorder, adjustment disorder, and other/unspecified reactions to severe stress), matched them by age and sex to individuals without any stress disorder diagnoses in the general population (1:5 ratio), and identified subsequent HPV-related cancer diagnoses (cervical, anal, vulvar, vaginal, penile, base of tongue, tonsillar, oropharyngeal) through 2022. Follow-up began after a 1-year lag period and was censored at first HPV-related cancer, other cancer diagnosis, emigration, death, or end of study. Covariates were measured prior to stress disorder diagnosis or matching date. We used Cox proportional hazards regression to estimate the sex-stratified adjusted hazard ratio (aHR) and 95% confidence interval (CI) of any HPV-related cancer and specific HPV-related cancers, adjusting for potential confounders.
During a median follow-up of approximately 10 years, there were 755 cases of HPV-related cancer in the stress disorder cohort (n=242,828) and 2,820 cases in the comparison cohort. Women with diagnosed stress disorders were more likely to develop any HPV-related cancer (aHR=1.3, 95% CI: 1.1–1.4), while men with stress disorders had an elevated rate of anal cancer specifically (aHR=1.7, 95% CI: 1.0–2.9) than their counterparts without stress disorders. The association between stress disorders and HPV-related cancer increased with younger age at stress disorder diagnosis and with longer follow-up time.
Our results suggest an association between stress disorders and slightly elevated rates of HPV-related cancer, which strengthens with longer follow-up, underscoring the need to understand biological and behavioral pathways linking stress disorders to increased risk of HPV-related cancer.
Stress disorders, including acute stress reaction, posttraumatic stress disorder, adjustment disorders, and other or unspecified stress reactions, have well-documented associations with a wide array of physical health consequences.1,2 Individuals with stress disorders experience physiological alterations including activation of the autonomic nervous system and the hypothalamic–pituitary–adrenal axis which results in the release of stress hormones including catecholamines and glucocorticoids.2–4 These stress hormones can be immunosuppressive and may explain increased susceptibility to some diseases.2 Furthermore, stress hormones exhibit effects on the tumor microenvironment including the promotion of tumor-cell growth and stimulation of angiogenesis suggesting the potential for increased cancer risk.4
Human papillomavirus (HPV) infection is a common sexually transmitted infection causally related to several anogenital and oropharyngeal precancers and cancers.5 Most HPV infections are asymptomatic and are cleared by the immune system within 12 to 24 months.6 However, infection may persist or become undetectable as latent or low-level chronic infections with risk of redetection,6 thus, increasing the likelihood of progression to precancerous lesions and HPV-related cancers.7,8 HPV-related cancers are largely preventable through HPV vaccination.9,10 In Denmark, prophylactic vaccines against HPV infection were implemented for administration to 12-year old girls in 2009 and to boys of the same age in 2019.11,12 Following initially high HPV vaccine uptake among eligible girls in 2009, uptake declined considerably from 2013–2016 in response to negative media coverage before a national campaign helped to restore uptake to baseline levels.12,13 HPV infection is more prevalent in men;14 however, HPV-related cancer is more common in women, largely driven by cervical cancer, a leading cause of cancer morbidity and death.5,8 Cervical cancer is the only HPV-related cancer for which screening is available; women ages 23–64 are invited for cervical cancer screening every 3–5 years.5,15 Cervical cancer screening participation was about 65% in 201016 and declined to 61% in 2022.17
A growing body of literature supports the theory that immune suppression is associated with an increased risk of redetection of HPV infection.6,18 Individuals with severe stress resulting in a stress disorder diagnosis may experience immune suppression, which lowers the ability to clear HPV infections and increases risk of HPV-related precancers and cancers.7,8 Yet, previous studies either report no association or weak associations between stress disorders19,20 or stressful life events (eg, bereavement) and cancer.21,22 One study from Sweden found that individuals experiencing bereavement of a close family member had an increased odds of HPV infection and subsequent cervical cancer.23 Another study reported that those experiencing the loss of a child had an increased risk of cervical cancer and oral cavity and pharyngeal cancers.24
Individuals with stress disorders may also be at increased risk of HPV infection and subsequent cancer progression through behavioral mechanisms. Population-based studies from Sweden found that women with a psychiatric diagnosis were 5% less likely to participate in cervical cancer screening compared to matched controls,25 and young girls with a psychiatric diagnosis were less likely to initiate the HPV vaccine series.26 Women with a psychiatric disorder who were misusing substances were nearly 20% less likely to participate in cervical cancer screening compared to women without psychiatric disorders, while those with a tobacco or alcohol-related disorder were 13% and 12% less likely to be screened, respectively.25 Individuals with mental illness are also more likely to have a higher number of sexual partners,27–29 a risk factor for HPV infection,30 and subsequent cancers including oropharyngeal31,32 and cervical cancer.33
Evidence of the association between stress disorders and HPV-related cancers may give insight into the physical health consequences of stress disorders and the possible role of immune suppression and behavioral mechanisms on the risk of HPV-related cancers. In this population-based cohort study, we examined sex-stratified associations between incident hospital-diagnosed stress disorders and the rate of HPV-related cancers in the population of Denmark from 1995 to 2022. This study addresses limitations of prior research on stress and HPV-related cancers by focusing on clinically diagnosed stress disorders rather than specific stressful life events.
Denmark has a population of approximately six million people.34 In Denmark, citizens benefit from a universal healthcare system with free access to both general practitioners and hospitals including preventative care (eg, HPV vaccination and cervical cancer screening) and treatment for mental disorders35 Healthcare and social data is tracked in nationwide registries, which can be linked using the civil registration number assigned to all Danish residents at birth or immigration.36 The base population for this study was Danish-born citizens aged ≥16 years.
We captured diagnoses of incident stress disorders using data from the Danish National Patient Registry, which covers all inpatient non-psychiatric hospital treatment in Denmark since 1977 and outpatient clinic and emergency room visits, including psychiatric diagnoses, since 1995.37,38 To ensure accuracy of incident stress disorder diagnoses, we also used the Danish Psychiatric Central Research Registry to identify stress disorder diagnoses from 1994. This registry is a record of inpatient psychiatric hospitalization information since 1969.39,40 Both registries have a high degree of completeness and in particular, the classification of stress disorders in the psychiatry registry has a high validity when compared with medical records.41 Details of the stress disorder cohort have been described elsewhere.42
For outcome ascertainment, we used the Danish Cancer Registry, which contains valid information on incident cancer diagnoses in Denmark since 1943.43 To obtain demographic information (ie, age, sex and marital status), we used the Danish Civil Registration System. This system records information on vital status and migration for the entire population in Denmark since 1968.36
We identified the stress disorder cohort using International Classification of Diseases, Tenth Revision (ICD-10) diagnostic codes (Table S1). Patients were included if they had a hospital-based stress disorder diagnosis of either acute stress reaction, posttraumatic stress disorder, adjustment disorder, other reactions to severe stress, or unspecified reactions to severe stress recorded in the Danish National Patient Registry between 1 January 1995 and 31 December 2021. Patients who received multiple stress disorder diagnoses were classified according to their first diagnosis.
For each member of the stress disorder cohort, we selected, through the Danish Civil Registration System,36 five Danish-born residents without a stress disorder diagnosis who were matched on age, sex, and date of the first diagnosis of an eligible stress disorder (index date). Follow-up time for members of the comparison cohort who were diagnosed with a stress disorder during follow-up was censored at the date of stress disorder diagnosis.
Using the Danish Cancer Registry, we identified all incident cancer diagnoses from 1 January 1995 to 31 December 2022. We followed patients until first HPV-related cancer (cervical, anal, vulvar, vaginal, penile, base of tongue, tonsillar, or oropharyngeal cancer) and follow-up was censored at date of any other cancer diagnosis, death, or 31 December 2022, whichever came first. Patients diagnosed with any cancer prior to follow-up and patients who died or were diagnosed with cancer within the first year of follow-up were excluded from the study. These exclusions were applied to mitigate reverse causation by ensuring the stress disorder preceded any cancer diagnosis and also limit early detection bias. HPV-related cancer sites were defined using ICD-10 codes; we could not determine whether cancers at these sites were attributable to HPV infection in our data.
We considered age at first diagnosis of a stress disorder, sex, marital status and comorbidity status as potential confounders, as these were factors believed to be associated with both stress disorders and cancer, based on existing literature review and expert medical knowledge.42,44–46 For comorbidity status, we computed a modified version (calculated without the cancer groups) of the Charlson Comorbidity Index (CCI)47,48 based on data from the Danish National Patient Registry, and we also examined specific CCI conditions separately. We obtained information on hospital-based diagnoses of other psychiatric illnesses including anxiety, depression, alcohol use disorder, and substance use disorder from the patient and psychiatry registries. Diagnoses from the patient registry were based on the International Classification of Diseases, Eighth Revision (ICD-8) before 1993 and ICD-10 thereafter, while all diagnostic codes from the psychiatry registry were based on the ICD-10 (Table S1).49 All covariates were assessed prior to first stress disorder diagnosis or matching date.
We examined the distribution of demographic characteristics and comorbidity level for individuals in the stress disorder and comparison cohorts. We calculated the median and interquartile range (IQR) of follow-up time and of age at first diagnosis of a stress disorder, and sample proportions for other characteristics and comorbidities. The absence of recorded diagnoses in the registry was treated as the absence of the condition and not as missing data. We followed patients from one year after the date of stress disorder diagnosis of the corresponding patient (index date) until first cancer diagnosis, date of death, emigration, or 31 December 2022, whichever came first.
We computed sex-stratified incidence rates (IR) per 10,000 person-years and 95% confidence intervals (CIs) of any HPV-related cancer and specific HPV-related cancer types in the stress disorder and comparison cohorts. Using Cox proportional hazards regression, we computed sex-stratified hazard ratios (HR) and 95% CIs of any HPV-related cancer and specific cancer types in patients with stress disorders compared to the matched comparison cohort. We adjusted for potential confounders including age, sex, and year of diagnosis by design and additionally adjusted for marital status, CCI score, anxiety, depression, alcohol use disorder and substance use disorder in multivariable models. We examined the assumption of proportional hazards graphically and determined the assumption to be appropriate. In secondary analyses, we stratified results by age at stress disorder diagnosis and length of follow-up. Furthermore, we restricted to women and stratified results by 1) stress disorder diagnoses (or matching) pre-2009 versus post-2009, when prophylactic vaccines against HPV infection were first introduced for 12-year old girls in Denmark,11 and 2) women in birth cohorts which were eligible vs. ineligible for HPV vaccination. Women eligible for the vaccine included those born in 1996 or later and those born in 1985–1992 with an index date after 1 August 2012, when vaccination catch-up programs were implemented.
To assess the robustness of our results, we conducted sensitivity analyses using alternative lag periods of 6 months and 2 years between stress disorder and HPV-related cancer diagnoses. We also estimated crude and adjusted hazard ratios across increasing follow-up durations. Both analyses were stratified by sex.
The study was reported to the Danish Data Protection Agency (approval 2016–051-000001/812). A completed STROBE checklist is included in supplementary material (Table S2). Figure 1Stress disorder and comparison cohort selection, Denmark, 1995–2022.The flowchart details the cohort selection for a stress disorder study. Starting with 5,783,507 Danish-born citizens aged 16+ from 1995-2021, 261,714 diagnosed with stress disorders (acute stress reaction, PTSD, adjustment disorder, unspecified severe stress reactions) from 1995-2021 are considered. After excluding 13,601 with prior cancer diagnoses and 5,285 who died or developed cancer within a year, 242,828 individuals remain, forming the stress cohort. A comparison cohort includes 1,240,565 individuals without stress disorders or prior cancer, matched by age, sex and diagnosis date. Exclusions for this group include 15,978 who died or developed cancer within a year and 25,189 broken matches. This forms the comparison cohort of 1,199,398 individuals.Flowchart of cohort selection for stress disorder study, showing exclusions and final cohorts.Note: Excluding broken matches means that if a patient with stress disorder is excluded, then we exclude their matched comparison as well. If all comparisons are excluded within the same stratum, then we exclude the patient with stress disorder as well; however, this did not occur in our sample.
Figure 1 displays the stress disorder and comparison cohort selection. Our cohort included 242,828 patients with a hospital-based diagnosis of a stress disorder with a median follow-up time of 10.2 years (IQR: 5.5–16.6 years). The comparison cohort included 1,199,398 individuals without a stress disorder diagnosis with a median follow-up time of 10.5 years (IQR: 5.7–17.0 years). The median age at index date was 36.8 years (IQR: 24.9–49.2 years) for the stress disorder cohort and 36.7 years (IQR: 24.9–49.0 years) for the comparison cohort. Demographic characteristics and comorbidities of the two cohorts are displayed in Table 1.Table 1Baseline Characteristics of the Two Study Cohorts, Restricted to Those Who are Both Cancer-Free and Alive 1 Year After Stress Disorder Diagnosis, Denmark, 1995 to 2022Stress DisorderCohort (N=242,828)ComparisonCohort (N=1,199,398)N (%)N (%)Length of follow-up time, years (median, IQR)10.2 (5.5–16.6)10.5 (5.7–17.0)Age at diagnosis, years (median, IQR)36.8 (24.9–49.2)36.7 (24.9–49.0)Specific stress-related disorders Acute stress reaction25,723 (10.6)127,139 (10.6) Posttraumatic stress disorder17,870 (7.4)88,436 (7.4) Adjustment disorder126,090 (51.9)622,809 (51.9) Other reaction to severe stress6,753 (2.8)33,379 (2.8) Reaction to severe stress, unspecified66,392 (27.3)327,635 (27.3)Sex Women145,183 (59.8)716,882 (59.8) Men97,645 (40.2)482,516 (40.2)Marital status Married/registered partnership72,983 (30.1)475,433 (39.6) Single85,483 (35.2)362,219 (30.2) Divorced35,035 (14.4)94,568 (7.9) Widow8,699 (3.6)28,823 (2.4) Unknown40,628 (16.7)238,355 (19.9)Charlson Comorbidity Index (CCI) score CCI score = 0196,978 (81.1)1,066,257 (88.9) CCI score ≥ 145,850 (18.9)133,141 (11.1)CCI specific comorbidity diagnoses Myocardial infarction3,291 (1.4)8,722 (0.7) Congestive heart failure2,050 (0.8)5,250 (0.4) Peripheral vascular disease2,935 (1.2)8,919 (0.7) Cerebrovascular disease7,246 (3.0)17,528 (1.5) Dementia736 (0.3)1,426 (0.1) Chronic pulmonary disease19,075 (7.9)56,572 (4.7) Connective tissue disease4,998 (2.1)17,084 (1.4) Ulcer disease5,120 (2.1)10,198 (0.9) Mild liver disease3,222 (1.3)5,102 (0.4) Diabetes I and II6,558 (2.7)17,948 (1.5) Hemiplegia699 (0.3)2,033 (0.2) Moderate to severe renal disease2,468 (1.0)7,129 (0.6) Diabetes with end organ damage2,875 (1.2)6,768 (0.6) Moderate to severe liver disease659 (0.3)1,009 (0.1) AIDS306 (0.1)515 (0.0)Other comorbidity diagnoses Anxiety disorder15,268 (6.3)13,420 (1.1) Depression37,164 (15.3)21,568 (1.8) Alcohol use disorder24,817 (10.2)28,384 (2.4) Substance use disorder13,981 (5.8)13,584 (1.1)Abbreviations: IQR, interquartile range; CCI, Charlson comorbidity index; AIDS, acquired immune deficiency syndrome.
The most common stress disorder diagnosis was adjustment disorder (51.9%) followed by unspecified reaction to severe stress (27.3%). A larger proportion of patients in the stress disorders cohort were single or divorced as compared to individuals in the comparison cohort (49.6% versus 38.1%), and a slightly higher proportion of patients with stress disorders were widowed (3.6% versus 2.4%). Patients with stress disorders had a higher level of comorbidity than individuals in the comparison cohort (CCI score ≥1: 18.9% versus 11.1%), particularly diagnoses of cerebrovascular disease, chronic pulmonary disease, and diabetes. The stress disorders cohort had a higher prevalence of hospital-based diagnoses of other psychiatric illnesses than the comparison cohort, including anxiety disorder (6.3% versus 1.1%), depression (15.3% versus 1.8%), alcohol use disorder (10.2% versus 2.4%) and substance use disorder (5.8% versus 1.1%; Table 1). Figure 2Rates of HPV-related cancer types among patients with stress disorders compared to a comparison cohort, stratified by sex, Denmark, 1995 to 2022.Image presents a two-part forest plot comparing hazard ratios for HPV-related cancers by sex, with Unadjusted and Adjusted models. For women, the x-axis shows Hazard Ratios (95% CI) from 0 to 3, with a reference line at 1. Cancers listed Any HPV-related (Unadjusted: 1.4; Adjusted: 1.3), Cervical (1.3; 1.2), Anal (1.3; 1.2), Vulvar (1.6; 1.3), Vaginal (1.3; 1.2), Base of tongue (1.4; 1.4), Tonsillar (1.6; 1.4) and Oropharyngeal (1.8; 1.1). For men, the same x-axis and reference line apply. Cancers listed Any HPV-related (1.3; 1.1), Anal (1.8; 1.7), Penile (0.97; 0.79), Base of tongue (1.4; 0.98), Tonsillar (1.3; 1.0) and Oropharyngeal (1.4; 1.2).A forest plot of hazard ratios for human papillomavirus related cancers by sex, comparing two models.Notes: Adjusted models are adjusted for cohabitation status, level of comorbidity, anxiety, depression and alcohol and drug dependency diagnoses.Abbreviations: HPV, human papillomavirus; CI, confidence interval.
Overall, there were 755 incident cases of HPV-related cancers among patients with any stress disorder over 2,774,467 person-years of follow-up. In the comparison cohort, there were 2,820 cases of HPV-related cancers over a total of 13,998,218 person-years of follow-up. Sex-stratified hazard ratios and 95% CIs are displayed in the Figure 2 forest plot, and the IRs per 10,000 person-years and 95% CIs are listed in Table S3. For both men and women, patients in the stress disorder cohort had a higher rate of any HPV-related cancer compared to the comparison cohort (men: IR = 1.9, 95% CI: 1.6–2.2 vs IR = 1.5, 95% CI: 1.4–1.6; IR = 3.7, 95% CI: 3.4–4.0 vs IR = 2.7, 95% CI: 2.6–2.8). Among women, we found an association between stress disorders and any HPV-related cancer after adjusting for confounders (adjusted HR (aHR) = 1.3, 95% CI: 1.1–1.4). Women with stress disorders experienced a higher rate of cervical cancer specifically, compared to their counterparts without stress disorders (IR = 2.2, 95% CI: 1.9–2.4 vs IR = 1.7, 95% CI: 1.6–1.8; aHR = 1.2, 95% CI: 1.1–1.4). We also found an elevated aHR of vulvar cancer (aHR = 1.3, 95% CI: 0.92–1.8), anal cancer (aHR = 1.2, 95% CI: 0.89–1.7) and tonsillar cancer (aHR = 1.4, 95% CI: 0.94–2.1) among women; however, the estimates were based on small numbers and the confidence intervals included 1. Among men, the aHR for any HPV-related cancer was 1.1 (95% CI: 0.91–1.3), and the rate of anal cancer was elevated in men with stress disorders compared to those without stress disorders (IR = 0.3, 95% CI: 0.2–0.4 vs IR = 0.2, 95% CI 0.1–0.2; aHR = 1.7, 95% CI: 1.0–2.9).
In stratified analyses, we found a stronger association between stress disorders and HPV-related cancers among patients diagnosed with a stress disorder at a younger age (16–39 years old at the time of diagnosis; aHR = 1.4, 95% CI: 1.2–1.6; Table 2) than among patients diagnosed with a stress disorder at age ≥ 40 years (aHR = 1.1, 95% CI: 0.96–1.2). Moreover, the HR of HPV-related cancer increased with increasing length of follow-up (aHR =1.3; 95% CI: 1.2–1.4 for ≥ 5 years of follow-up versus 1.0; 95% CI: 0.84–1.2 for 1–4 years of follow-up). Among women diagnosed with stress disorders before 2009, there was a stronger association with HPV-related cancers (aHR = 1.3, 95% CI: 1.2–1.5) than there was among women diagnosed in 2009 or later (aHR = 1.0, 95% CI: 0.84–1.3). About 20% of women belonged to birth cohorts eligible for HPV vaccination. We found an association between stress disorders and rate of any HPV-related cancer among women ineligible for HPV vaccination (aHR = 1.2, 95% CI: 1.1–1.4). Among those eligible, the aHR was stronger yet imprecise (aHR = 1.4, 95% CI: 0.73–2.6).Table 2Rate of Any HPV-Related Cancer Among Patients with Stress Disorders Compared to a Comparison Cohort, Stratified by Variables of Interest, Denmark, 1995 to 2022Stratification VariableStress Disorder CohortComparison CohortUnadjusted ModelModel Adjusted forMarital Status, CCI andPsychiatric DisordersNTotalPerson-YearsIR per 10,000Person-Years(95% CI)NTotalPerson-YearsIR per 10,000Person-Years(95% CI)HR(95% CI)aHR (95% CI)Age at diagnosis, years16–393371,552,0262.2 (1.9–2.4)11247,658,5971.5 (1.4–1.6)1.5 (1.3–1.7)1.4 (1.2–1.6)≥40418979,6134.3 (3.9–4.7)16965,140,2233.3 (3.1–3.5)1.3 (1.2–1.4)1.1 (0.96–1.2)Length of follow-up, years1–4 years173865,9352.0 (1.7–2.3)7444,301,4921.7 (1.6–1.9)1.1 (0.97–1.4)1.0 (0.84–1.2)≥5 years5821,665,7053.5 (3.2–3.8)20768,497,3281.5 (1.3–1.6)1.5 (1.3–1.6)1.3 (1.2–1.4)Year of diagnosis^a^Before 20094381,050,5203.9 (3.6–4.3)15155,268,3152.9 (2.7–3.0)1.5 (1.3–1.7)1.3 (1.2–1.5)2009 or later129492,5482.2 (1.9–2.6)5492,418,1562.3 (2.1–2.5)1.1 (0.93–1.4)1.0 (0.84–1.3)HPV vaccination eligibility^a,b^Eligible16136,8911.2 (0.6–1.7)50665,7410.8 (0.5–1.0)1.5 (0.87–2.7)1.4 (0.73–2.6)Ineligible5511,406,1773.9 (3.6–4.2)20147,020,7302.9 (2.7–3.0)1.4 (1.3–1.5)1.2 (1.1–1.4)Notes: Incidence rates are presented per 10,000 person-years. All hazard ratios adjusted for age, sex and calendar year by design. Adjusted HRs are additionally adjusted for marital status, Charlson Comorbidity Index (CCI) score, anxiety, depression and alcohol and substance use disorder diagnoses. ^a^ Restricted to women. ^b^ Women eligible for the HPV vaccine included those born in 1996 or later and those born in 1985–1992 with an index date after 1 August 2012.Abbreviations: HPV, human papillomavirus; CCI, Charlson Comorbidity Index; IR, incidence rate; HR, hazard ratio; aHR, adjusted hazard ratio; CI, confidence interval.
In sensitivity analyses, we found no meaningful differences with a 6-month or 2-year lag period, with the exception of anal cancer among men, for which the hazard ratios were attenuated with a 2-year lag period (aHR = 1.4, 95% CI: 0.84–2.5; data not shown). With increasing durations of follow-up, aHRs for the association between stress disorders and any HPV-related cancer increased among women (Table S4). Among men, aHRs were initially high from 1–3 years follow-up, declined with 1–5 years follow-up, and then increased slightly with follow-up periods of 10 years or greater; however, confidence intervals were imprecise and contained the null value of 1.
In this nationwide cohort study from 1995 to 2022, we observed a slightly elevated relative rate of HPV-related cancers among women with a stress-disorder diagnosis compared to matched controls from the general population. The association between stress disorders and HPV-related cancer increased with longer follow-up time and was stronger for individuals diagnosed with stress disorders at a younger age (the former of which is likely consistent with the expected latency period of stress disorder diagnosis to HPV-related cancer onset).
Our findings are consistent with previous studies which found a slightly increased risk of cancer in patients with stress disorders19,20 or among individuals experiencing stressful life events.21,22 A Swedish cohort study by Fang et al24 showed an increased risk of cervical cancer, vulvar and vaginal cancer, anal cancer and oral cavity and pharyngeal cancers in individuals experiencing bereavement from the loss of a child. Similarly, another Swedish study by Lu et al23 showed that individuals experiencing the loss of a child, spouse or sibling had increased odds of HPV infection and cervical cancer. Our results are slightly weaker in magnitude from those reported by Fang et al and Lu et al, possibly because we included additional adjustment for level of physical comorbidity and psychiatric health. Another difference between our study and this extant work is that we focused on stress disorders broadly rather than a specific stressful or traumatic event. Stress disorder diagnoses can be based on a wide variety of experiences that range in severity, and this could explain differences in the magnitude of results. This study adds to this literature by specifically examining sex-stratified associations between diagnosed stress disorders and HPV-related cancers. Taken together, this body of evidence indicates that stressful and traumatic life experiences, as well as stress disorders, increase risk for some HPV-related cancers.
Our sex-stratified analyses showed that stress disorders were associated with increased rates of HPV-related cancers among women, but not among men. When examining specific cancer sites, stress disorders were most strongly associated with cervical cancer in women and anal cancer in men. As these are the most common cancer sites, these findings likely reflect variation in outcome composition and statistical power. However, biological factors, differences in screening participation, and other behavioral risk factors may also contribute to the observed sex differences. Future studies are needed to better understand how the mechanisms linking stress disorders and HPV-related cancers are modified by sex.
In our study population, the median age at stress disorder diagnosis (or matching) was 37 years, with a median follow-up time of 10 years. For most HPV-related cancers, the median age at diagnosis is above 60 years, with the exception of cervical cancer, for which the median age at diagnosis is 50 years.50,51 Therefore, our study period was likely too short to capture all cases of HPV-related cancer following a stress disorder diagnosis, as most individuals in the cohort did not reach the median age of cancer diagnosis during follow up. This is supported by our finding that the association between stress disorders and HPV-related cancer increased with longer follow-up time, suggesting that extended follow-up allows for more complete capture of HPV-related cancer cases and a more accurate estimate of the association.
We found that women diagnosed with a stress disorder before 2009, when HPV vaccines were first administered to 12-year-old girls in Denmark, had elevated relative rate of HPV-related cancers than women diagnosed with stress disorders after 2009. This finding most likely merely reflects increased length of follow-up time, as most women in this study were not eligible for HPV vaccination. Among women in non-eligible birth cohorts, we observed stronger evidence of an association between stress disorders and HPV-related cancers, which may also reflect longer follow-up time or the direct effect of HPV vaccination. In Denmark, rates of HPV-related cancers have remained stable or increased slightly since 2009, with the exception of cervical cancer, which has been decreasing since 2015 (Figure 3).44,50,51 Therefore, it is also possible that the null association we observe between stress disorders and HPV-related cancers among women post-2009 is due to decreasing rates of cervical cancer, the cancer site with the most cases and strongest observed association with stress disorders in our study. For cancers other than cervical cancer with longer latency periods, follow-up may have been insufficient to observe potential effects of HPV vaccination. Figure 3Age-standardized incidence rates for HPV-related cancers and timing of HPV vaccination programs stratified by sex, Denmark, 1995 to 2022.Two line graphs show age-standardized cancer incidence rates in Denmark per 100,000 person-years from 1995 to 2022. For Women, the graph tracks Anal, Cervical, Oropharyngeal, Vaginal and Vulvar cancers. Cervical cancer decreases from 18 in 1995 to about 12 by 2022, with a low near 10 in 2021. Oropharyngeal cancer rises from 1.5 to 4 and Anal cancer from 1.5 to 3.5. Vulvar cancer fluctuates between 2.5 and 4, ending near 3.5 to 4. Vaginal cancer stays between 0.5 and 1.2, ending near 0.5. A dashed line appears around 2009, with shaded bands around 2012-2013. For Men, the graph includes Anal, Oropharyngeal and Penile cancers. Oropharyngeal cancer increases from 4.5 to 10-11, peaking around 2021. Penile cancer stays around 2 to 3, ending near 2.5. Anal cancer remains between 1 and 2, ending near 1.5 to 2. A dashed line is near 2019, with a shaded band around 2020-2021. Legends indicate HPV vaccine introduction and catch-up programs for specific cohorts.A multi-line graph showing HPV-related cancer incidence rates in Denmark from 1995 to 2022 by sex.Notes: Incidence data was retrieved from the NORDCAN database50,51 and uses the Nordic population in 2000 as the reference population. Information on timing of vaccine introduction and catch-up programs retrieved from Sander et al,11 Statens Serum Institut,12,52 and Bigaard et al12 a Vaccine introduction age corresponds to birth cohorts starting in 1996 for women and 2007 for men.Abbreviation: HPV, human papillomavirus.
Among men, we observed an association between stress disorders and anal cancer; however, this association was attenuated with a 2-year lag period. This could suggest that reverse causation may have influenced estimates using a 1-year lag period, such that stress disorders resulting from anal cancer symptoms or treatment may still have been captured. This finding may also relate to our observation that associations between stress disorders and any HPV-related cancer among men were stronger during the first 5 years of follow up, particularly if the overall association was driven largely by anal cancer cases. These patterns may reflect increased surveillance and earlier detection following a stress disorder diagnosis, followed by a compensatory decrease in incidence. However, we cannot make any firm conclusions, as these estimates were based on small numbers and imprecise.
Key strengths of this cohort study include use of population-based data over a long time period, obtained from nationwide registries with high validity. Our large sample enabled us to examine specific cancer sites with low frequency of incident cases and stratify results by important subgroups, including by sex and age.
Our findings could be substantially impacted by residual confounding, as we were unable to adjust for key confounders including lifestyle factors and behavioral characteristics, education level, body mass index, and sexual behavior. These unmeasured confounders are likely strongly linked to both stress disorders and HPV-related cancer risk, and their omission may meaningfully bias our estimates. For example, individuals with stress disorders may have a higher number of sexual partners, increasing exposure to HPV and potentially contributing to the observed associations with HPV-related cancers. We expect that adjusting for these characteristics may attenuate our findings; future research should adjust for these factors to determine whether our observed association persists. Additionally, our study population was comprised of patients with a hospital diagnosis of a stress disorder. Thus, our cohort did not capture patients diagnosed by their general practitioner or individuals for whom a stress disorder is undiagnosed, limiting generalizability of our findings. Our results likely represent associations between stress disorders with moderate to high severity and HPV-related cancers; mild cases of stress disorders may have different associations with HPV-related cancers, and this is an interesting area for further inquiry. Patients with more severe stress disorders may benefit from enhanced surveillance and clinical follow-up to prevent or detect subsequent cancers earlier. In an effort to limit reverse causation as well as early detection bias, patients diagnosed with any cancer prior to follow-up and patients who died or were diagnosed with cancer within the first year of follow-up were excluded from the study; however, this may have introduced some selection bias. Given that stress disorder diagnoses were hospital-based, surveillance bias is possible, leading to higher probability of HPV-related cancer diagnoses among those with stress disorders due to greater engagement with Denmark’s health care services. However, in Denmark, clinical follow-up of people diagnosed with stress disorders is most intensive immediately after diagnosis – the stronger association found with increased follow-up time suggests that surveillance bias could not fully account for our results. Finally, despite the long follow-up period, it may have been insufficient to capture all cases of HPV-related cancer, especially those with prolonged latency that occur at older ages.
We found a slightly increased relative rate of HPV-related cancers among women with stress disorder diagnoses, driven primarily by cervical cancer, as well as anal cancer among men, and these associations strengthened with longer follow-up time. Our findings highlight the potential of early psychological interventions to indirectly reduce HPV-related cancer risk, as well as the need for additional research on biological and behavioral pathways through which stress disorders may increase risk of subsequent HPV-related cancers. Individuals with stress disorders may represent a population which would particularly benefit from targeted HPV screening and vaccination programs.