Background
A cancer diagnosis often comes with considerable emotional and psychological burden. Among the most serious sequelae are suicide events, which are fortunately rare. Nonfatal self-injury (NFSI) events, whether conceptualized as suicide attempts or self-injury to manage ongoing severe distress and difficulty coping, are more common.1,2 The cumulative incidence of NFSI following a cancer diagnosis has been reported as significantly higher than that of suicide (0.3% vs <0.1%), with the risk peaking among those with a history of self-injury (10.2%).3
National and international cancer agencies recommend screening for and managing distress that might indicate a mental health disorder in patients with cancer, from diagnosis through survivorship.4,5 Despite these recommendations, few patients with cancer-related distress receive support.6 Although most patients who undergo specialized assessment and intervention show improvement, few patients with cancer-related distress receive such support.6 Greater focus and resources are needed to address this gap, which may contribute to later serious sequelae, such as NFSI and suicide.
Although NFSI is an adverse outcome in and of itself, there is a lack of evidence regarding its implications for patients already facing considerable morbidity, such as those newly diagnosed with cancer. Better understanding the burden that an event such as NFSI places on patients with cancer and health care systems could help drive efforts to prioritize its prevention, inform resource allocation and investment, and guide the design of supportive care pathways in oncology. For example, although NFSI can lead to reduced all-cause life expectancy in the general population,7 it is unknown whether the same applies to patients with cancer. This study, therefore, examined the association between NFSI events following a cancer diagnosis and survival outcomes in this population.
Methods
Study Design
A retrospective population-based cohort study was conducted. Administrative datasets were linked using unique encoded identifiers and analyzed at ICES. The study was approved by the Research Ethics Board at Sunnybrook Health Sciences Centre and was reported in accordance with the REporting of studies Conducted using Observational Routinely collected health Data (RECORD) statement.8 A cancer survivor (J. Deleemans) was part of the research team, contributing to the development of the research question, planning and conducting the analysis, and interpreting the results. This involvement brought the perspective of lived cancer experience and ensured the findings were meaningful and relevant to patients.
Data Sources
The Ontario Cancer Registry (OCR) is a provincial database that includes nearly all (96%) patients diagnosed with cancer (excluding nonmelanoma skin cancers) since 1964.9,10 The Registered Persons Database (RPDB) provides vital status and demographic data for all individuals covered under the Ontario Health Insurance Plan (OHIP). Information regarding health services is captured in several sources, including the Canadian Institute of Health Information Discharge Abstract Database (CIHI-DAD), the National Ambulatory Care Reporting System (NACRS), Cancer Activity Level Reporting (ALR), the OHIP Claims Database, and the Ontario Mental Health Reporting System (OMHRS).11 Detailed descriptions of these datasets are available in Table S1 (available online in the supplementary materials).
Study Cohort
Under the Canada Health Act, Ontario’s 14.5 million residents benefit from universally accessible and publicly funded health care through OHIP.12 As described in our previous study, adults (age ≥18 years) with a diagnosis of cancer between January 1, 2007, and March 31, 2019, were identified using ICD-10-O.3 codes (Supplementary Table S2) from the OCR.3 If more than one cancer diagnosis existed within the study window, the earlier date was selected. Patients were excluded if they were aged >105 years, they had >1 cancer diagnosis on the same day, their date of last contact was missing, or they died prior to their documented date of cancer diagnosis.
Exposure
The exposure of interest was an NFSI event at any time after cancer diagnosis. NFSI was defined as an emergency department visit where a self-injury of intentional (ICD 10-CA codes X61–X84) or undetermined intent (ICD-10-CA codes Y10–Y19, Y28) was recorded in any diagnostic field, as previously described.3,13–15 Given that the time between cancer diagnosis and NFSI may vary among patients, NFSI was treated as a time-dependent exposure variable, updated in instantaneous time.16
Outcome
The primary outcome was overall survival (OS), defined as time to death from any cause measured from the time of cancer diagnosis. Patients were followed until date of death, date of last contact, or end of study date on June 30, 2021, whichever occurred first. Patients whose follow-up ended due to loss of contact or end of study were censored at that time.
Covariates
Patient age and sex were captured. Rural living was determined using the Rurality Index of Ontario, with index ≥40 representing rural residence.17 Socioeconomic status (SES) was captured using the Material Deprivation Index, a composite index of the inability for individuals or households to afford consumption goods and activities typical in a society at a given point in time, categorized into quintiles.18,19 Comorbidity burden was measured using Elixhauser’s algorithm with a 2-year lookback window, excluding the present cancer diagnosis and any historical cancer diagnoses during the lookback period.20 The total comorbidity score was dichotomized, with a cutoff of ≥4 indicating high comorbidity burden.21,22 Cancer site and stage at the time of diagnosis according to the seventh and eighth editions of the AJCC staging system were abstracted from the OCR.23 Psychiatric illness in the 5 years prior to cancer diagnosis was categorized into 4 mutually exclusive groups representing levels of severity of mental health care needs, as previously described24: (1) Inpatient severe psychiatric illness: any hospitalization for a mood or psychotic disorder, (2) Outpatient severe psychiatric illness: >2 psychiatry outpatient visits with or an emergency department visit for a mood or psychotic disorder; (3) Other mental illness: <2 psychiatry outpatient visits with any physicians (such as psychiatrist or family physician) or emergency department visit with any mental illness diagnosis codes; and (4) None: no history of mental health services use. Prior self-injury was captured in the 5 years prior to cancer diagnosis and treated as a dichotomous yes/no variable. Finally, cancer treatment received was categorized as none, surgery alone, chemo(radio)therapy alone, or surgery followed by adjuvant therapy (chemotherapy or chemoradiation/radiation therapy).
All covariates, except for prior self-injury, prior psychiatric illness, cancer site, and cancer stage at diagnosis, were measured at the time of cancer diagnosis and updated annually to reflect patient and clinical characteristics before the NFSI exposure event. Thus, the covariates adjusted for in the regression models represent their value around the time of the NFSI event.
Statistical Analysis
Characteristics of the cohort at the time of cancer diagnosis were first examined for the entire cohort and then stratified by the presence or absence of NFSI for descriptive purposes. Categorical variables were reported as absolute number (n) and proportion (%), and continuous variables as median with IQR. Chi-square and Mann-Whitney tests were used to compare the distribution of categorical and continuous variables between the 2 groups.
Due to the time-varying nature of the primary exposure (NFSI), Kaplan-Meier methods and the log-rank test were not appropriate for examining survival probability. Instead, the time-varying nature of the exposure was accounted for using an extended Kaplan-Meier estimator, the Snapinn technique, to assess differences in survival outcomes between the 2 groups.25 The association between NFSI and all-cause mortality was examined using an extended Cox multivariable regression model.16 Covariates were identified a priori as potential confounders to adjust for and include in the models based on clinical relevance and the existing literature, using directed acyclic graphs.26 These covariates included age at diagnosis, sex, comorbidity burden, SES, rural residence, prior self-injury, history of psychiatric illness, primary cancer site, and cancer therapy received. Results were reported as hazard ratios (HRs) with 95% confidence intervals. Under a time-dependent framework, this analysis estimates the relative hazards of death for patients with NFSI compared with those without NFSI, considering the time elapsed since their cancer diagnosis. The NFSI exposure variable was treated as (1) dichotomous (yes vs no), where the NFSI variable “turned on” at the first NFSI event; and (2) count-based, where the risk associated with each additional NFSI event was examined in the case of multiple events per patient.
Missing data were assessed for each variable. Rural residence and SES had missing data in 0.1% and 0.8% of cases, respectively. For these variables, a complete-case analysis was used in multivariable models. Stage at diagnosis had missing data in 38.2% of cases. Because missing data may indicate a lack of measurement and is not random, a separate “missing” category was created for stage. All analyses were 2-sided, with statistical significance set at P<.05. Analyses were performed using SAS Enterprise Guide 7.1 (SAS Institute Inc.).
Results
Among 806,910 patients with a new cancer diagnosis during the study period, 2,482 (0.31%) experienced an NFSI event after their diagnosis, with a median time to the first NFSI event of 29 months (IQR, 11–57). Of those, 81 (3.2% of patients with an NFSI) experienced >1 NFSI event. Median follow-up for the entire cohort was 36 months (IQR, 12–84), during which 351,413 (43.6%) patients died. NFSI rates peaked at 7 per 10,000 patients within the first 6 months after diagnosis, with a mean 6-month NFSI rate of 5 per 10,000 patients over the first 5 years after diagnosis.
At the time of cancer diagnosis (Table 1), patients who would later experience an NFSI were younger, were more likely to be female, had higher comorbidity burden but lower SES, were more likely to have had a prior self-injury event or a history of psychiatric illnesses, and had an earlier rather than a more advanced cancer stage. The distribution of cancer type also differed between the 2 groups. Most patients received cancer-directed treatment within 1 year of diagnosis. Among those who experienced an NFSI, most underwent surgery within the first year, whereas most patients who did not experience an NFSI received no cancer-directed therapy during this period.
Patient Characteristics at the Time of Cancer Diagnosisa
| Characteristic | All Patients n (%) |
No NFSI n (%) |
NFSI n (%) |
P Value |
|---|---|---|---|---|
| Total, n | 806,910 | 804,428 | 2,482 | |
| Age at diagnosis, median (IQR), y | 67 (57–76) | 67 (57–76) | 57 (47–67) | <.01 |
| Sex | <.01 | |||
| Female | 401,749 (49.8) | 400,420 (49.8) | 1,329 (53.5) | |
| Male | 405,161 (50.2) | 404,008 (50.2) | 1,153 (46.5) | |
| High comorbidity burden (Elixhauser >4) | 62,347 (7.7) | 62,127 (7.7) | 220 (8.9) | .03 |
| Rural residence | 81,571 (10.1) | 81,312 (10.1) | 259 (10.4) | .59 |
| Socioeconomic status | <.01 | |||
| First (highest) | 166,794 (20.7) | 166,389 (20.7) | 405 (16.3) | |
| Second | 160,700 (19.9) | 160,276 (19.9) | 424 (17.1) | |
| Third | 159,064 (19.7) | 158,618 (19.7) | 446 (18.0) | |
| Fourth | 159,775 (19.8) | 159,246 (19.8) | 529 (21.3) | |
| Fifth (lowest) | 160,577 (19.9) | 159,899 (19.9) | 678 (27.3) | |
| Prior self-injury (in 5 years prior to cancer diagnosis) | 2,807 (0.3) | 2,555 (0.3) | 252 (10.2) | <.01 |
| History of psychiatric illness (in 5 years prior to cancer diagnosis) | <.01 | |||
| None | 489,951 (60.7) | 489,168 (60.8) | 783 (31.5) | |
| Inpatient severe psychiatric illness | 6,590 (0.8) | 6,286 (0.8) | 304 (12.2) | |
| Outpatient severe psychiatric illness | 14,257 (1.8) | 14,043 (1.7) | 214 (8.6) | |
| Other mental illness | 296,112 (36.7) | 294,931 (36.7) | 1,181 (47.6) | |
| Primary cancer site | <.01 | |||
| Bone/Sarcoma/PNS | 1,659 (0.2) | 1,654 (0.2) | 5 (0.2) | |
| Breast | 112,943 (14.0) | 112,536 (14.0) | 407 (16.4) | |
| Bronchopulmonary | 104,145 (12.9) | 103,891 (12.9) | 254 (10.2) | |
| Central nervous system | 10,669 (1.3) | 10,623 (1.3) | 46 (1.9) | |
| Endocrine | 32,335 (4.0) | 32,206 (4.0) | 129 (5.2) | |
| Gastrointestinal | 152,083 (18.8) | 151,716 (18.9) | 367 (14.8) | |
| Genitourinary | 170,475 (21.1) | 169,973 (21.1) | 502 (20.2) | |
| Gynecologic | 49,672 (6.2) | 49,496 (6.2) | 176 (7.1) | |
| Hematopoietic/Lymphoma | 92,635 (11.5) | 92,324 (11.5) | 311 (12.5) | |
| Head and neck | 19,220 (2.4) | 19,127 (2.4) | 93 (3.7) | |
| Skin | 36,842 (4.6) | 36,733 (4.6) | 109 (4.4) | |
| Other | 24,232 (3.0) | 24,149 (3.0) | 83 (3.3) | |
| Cancer stage | <.01 | |||
| 0/Ib | 151,397 (18.7) | 150,831 (18.7) | 566 (22.8) | |
| II | 144,590 (17.9) | 144,121 (17.9) | 469 (18.9) | |
| III | 94,262 (11.7) | 93,947 (11.7) | 315 (12.7) | |
| IV | 108,191 (13.4) | 107,999 (13.4) | 192 (7.7) | |
| Missing | 308,470 (38.2) | 307,530 (38.2) | 940 (37.9) | |
| Cancer-directed treatment (12 months after cancer diagnosis) | <.01 | |||
| None | 245,947 (30.5) | 245,299 (30.5) | 648 (26.1) | |
| Surgery alone | 221,523 (27.5) | 220,704 (27.4) | 819 (33.0) | |
| Chemo(radio)therapy alone | 164,409 (20.4) | 164,003 (20.4) | 406 (16.4) | |
| Surgery + chemo(radio)therapy | 175,031 (21.7) | 174,422 (21.7) | 609 (24.5) | |
Abbreviations: NFSI, nonfatal self-injury; PNS, peripheral nervous system.
Characteristics at the time of cancer diagnosis are presented for the purpose of cohort description; the characteristics adjusted for in the multivariable analyses were updated annually to reflect the values at the time of the exposure event of interest.
Stages 0 and I combined due to small cell report.
Figure 1 presents OS stratified by time-dependent NFSI status using extended Kaplan-Meier methods. In unadjusted analysis, patients who had experienced an NFSI at any given time had inferior OS compared with those who had not yet experienced an NFSI (HR, 1.73; 95% CI, 1.61–1.85). This association persisted after adjusting for relevant covariates (HR, 1.46; 95% CI, 1.30–1.62) (Supplementary Table S3). These findings indicate that, at any given time after a cancer diagnosis, patients who had experienced an NFSI had a 46% higher hazard of all-cause mortality compared with those who had not. When treating NFSI as a count variable, each additional NFSI event was associated with inferior OS in both unadjusted (HR, 1.19; 95% CI, 1.11–1.28) and adjusted (HR, 1.17; 95% CI, 1.08–1.27) analyses (Supplementary Table S4).
Probability of death from any cause stratified by NFSI status (time-dependent variable), using extended Kaplan-Meier methods.
Abbreviation: NFSI, nonfatal self-injury.
Citation: Journal of the National Comprehensive Cancer Network 23, 3; 10.6004/jnccn.2024.7083
Discussion
In this population-based cohort of 806,910 patients, experiencing an NFSI after a cancer diagnosis was associated with a 46% higher hazard of all-cause mortality compared with not yet experiencing an NFSI. Among patients with multiple NFSI events, each additional NFSI event was associated with a 17% increase in adjusted mortality hazard. These findings underscore the critical impact of NFSI on survival, highlighting the need for enhanced risk assessment and targeted interventions to reduce their occurrence and improve patients’ experiences and outcomes.
Previous studies have focused on depression and anxiety and identified their association with higher cancer-specific mortality and poorer cancer survival.27 Major depression has been linked to a 41% increase in mortality hazards among patients with cancer, consistent with the 46% increase observed in this study for patients with NFSI.28 Given that NFSI represents a serious mental health sequelae, this finding is not surprising. The current results provide novel insights into the relationships between high distress, associated mental health repercussions, and cancer outcomes. The findings build momentum and support the impetus for addressing NFSI as a critical postdiagnosis event. Patients with cancer have an adjusted 1.21-fold increased risk of NFSI (95% CI, 1.06–1.39) compared with matched controls.29 However, suicide rates in patients with cancer are not higher than those in noncancer controls (8.2 vs 11.4 per 1000 person-years), and both groups have similar cumulative incidences of suicide mortality.24 This suggests that the observed increased mortality risk associated with NFSI is unlikely to be attributable solely to suicide. Furthermore, half of the patients who experienced NFSI in this cohort reported it within 2.5 years of their cancer diagnosis. Such prolonged period of distress leading up to an NFSI event is concerning, especially considering that psychosocial screenings and support should typically be provided alongside cancer treatment, including during survivorship. As such, these results suggest that the burden of NFSI in patients with cancer has been overlooked and undertreated. Beyond being a significant burden for patients and caregivers, NFSI and its repercussions on outcomes also represent a considerable burden for clinicians and health systems. Addressing NFSI should be a priority throughout the cancer care continuum, with resources dedicated to active screening, early identification, and proactive treatment. Furthermore, NFSI should be considered a key outcome for assessment and monitoring of cancer care.
Mechanistically, the association between NFSI and survival is complex and likely multifactorial. Prior work has demonstrated that patients with preexisting mental health issues have decreased staging documentation and lower receipt of guideline-recommended treatment,30,31 both of which can adversely affect cancer outcomes. Similar mechanisms may be implicated in patients with NFSI. Additionally, underlying risk factors of NFSI, such as substance use, certain personality traits (eg, impulsivity and aggression), and low engagement in health promotion, may also negatively impact survival in patients with cancer. These factors are further intertwined with social determinants of health, including income, age, and sex.32 However, this study does not establish or suggest causality between NFSI and survival. Instead, NFSI serves as an important marker and potential measurement of distress following a cancer diagnosis. Its association with inferior survival outlines the need for enhanced patient support, perhaps by identifying those at greater risk. Better mental health management has previously been associated with improved symptom control in patients with cancer.6 Improved care pathways could therefore reduce NFSI and mitigate its repercussions for patients, caregivers, clinicians, and health care systems. For instance, NFSI is frequently not followed by outpatient assessment and follow-up.33 Additional work is needed to understand the complex interplay of socioeconomic, clinical, and psychosocial factors influencing NFSI in order to derive the most appropriate pathways, which was beyond the scope of the current work.
Two main areas could be targeted for quality improvement in patients diagnosed with cancer to better address NFSI. The first is identifying patients at high risk for severe distress, such as manifested by NFSI. This can be achieved through individual risk assessment and monitoring of patient-related outcomes, such as elevated depression symptoms.3,6,34,35 Factors associated with a higher risk of NFSI after cancer diagnosis have previously been described (and confirmed in the current analysis), including younger age, a history of psychiatric illness or self-injury, certain cancer types (head and neck, genitourinary, hematopoietic, and lymphoma), and earlier cancer stage.3 These factors should be documented as part of oncology assessments and may eventually be further supported by predictive scoring systems that integrate patient and disease characteristics. The second area for improvement involves providing timely, timely, integrated follow-up care, including mental health screening and treatment, for patients identified as being at high risk for severe distress or those already experiencing NFSI. Indeed, evidence of more severe patient-reported symptoms on screening has been associated with higher risk of NFSI after a cancer diagnosis. Reporting of moderate to severe anxiety and depression as well as a high total score on the Edmonton Symptom Assessment System are associated with higher odds of NFSI within 180 days after a cancer diagnosis.34 It is possible that unidentified or untreated mental health disorders contribute to the association between NFSI and mortality, suggesting that appropriate support and treatment could mitigate both. Improvement in identifying patients at risk and managing distress and its repercussions both require screening, response pathways, and specialized supportive care resources to address mental health during the cancer journey. However, specific clinical recommendations for interventions on patient-reported distress have not yet been defined, including establishing a threshold of symptom severity that warrants a clinical action, identifying the appropriate health care provider to respond to reported symptoms, and determining how soon to intervene.36–39
This study has several limitations. Due to its retrospective design, there is potential for bias, misclassification, and unmeasured confounding. For example, confounders such as substance use, impulsivity/sensation-seeking, and social support could not be captured, although physician visits related to substance abuse were included in the Elixhauser index. Additionally, the use of health administrative databases to identify NFSI events may have led to underestimation, as individuals who experienced an NFSI but did not seek formal help could have been missed, potentially biasing results toward the mean. Suicide was not included in the capture of NFSI, as previously published.3 This exclusion was deliberate, given that suicide is a rare event, challenging to predict, and represents a distinct sequela of mental illness compared with NFSI. Finally, this analysis deliberately focused on testing for an association between NFSI and mortality. Identifying the underlying mechanisms of this association fell beyond its scope. Future studies employing different designs will be necessary to provide insights into risk identification and potential intervention. Despite these limitations, this study employed rigorous methodology, including accounting for the time-varying nature of the exposure, excellent long-term follow-up, a large population-based sample, and minimal loss to follow-up. These strengths contributed to a robust evaluation of the association of NFSI with survival after a cancer diagnosis.
Conclusions
NFSI following a cancer diagnosis was associated with inferior survival, with patients experiencing a 46% higher mortality hazard compared with those who had not yet experienced NFSI at any given time. This highlights NFSI as a crucial event for patients with cancer that should be addressed proactively. Future research is needed to better understand the mechanisms underlying inferior survival in patients with NFSI and to design tailored, integrated supportive care pathways to address this issue effectively.
References
- 2.↑
Geulayov G, Casey D, McDonald KC, et al. Incidence of suicide, hospital-presenting non-fatal self-harm, and community-occurring non-fatal self-harm in adolescents in England (the iceberg model of self-harm): a retrospective study. Lancet Psychiatry 2018;5:167–174.
- 3.↑
Noel CW, Eskander A, Sutradhar R, et al. Incidence of and factors associated with nonfatal self-injury after a cancer diagnosis in Ontario, Canada. JAMA Netw Open 2021;4:e2126822.
- 4.↑
Holland JC, Bultz BD. The NCCN Guideline for distress management: a case for making distress the sixth vital sign. J Natl Compr Canc Netw 2007;5:3–7.
- 5.↑
Bultz BD, Groff SL, Fitch M, et al. Implementing screening for distress, the 6th vital sign: a Canadian strategy for changing practice. Psychooncology 2011;20:463–469.
- 6.↑
Hallet J, Davis LE, Isenberg-Grzeda E, et al. Gaps in the management of depression symptoms following cancer diagnosis: a population-based analysis of prospective patient-reported outcomes. Oncologist 2020;25:e1098–1108.
- 7.↑
Bergen H, Hawton K, Waters K, et al. Premature death after self-harm: a multicentre cohort study. Lancet 2012;380:1568–1574.
- 8.↑
Benchimol EI, Smeeth L, Guttmann A, et al. The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) statement. PLoS Med 2015;12:e1001885.
- 9.↑
Robles SC, Marrett LD, Clarke EA, Risch HA. An application of capture-recapture methods to the estimation of completeness of cancer registration. J Clin Epidemiol 1988;41:495–501.
- 10.↑
Clarke EA, Marrett LD, Kreiger N. Cancer registration in Ontario: a computer approach. IARC Sci Publ 1991:246–257.
- 11.↑
Juurlink D, Preyra C, Croxford R, et al. Canadian Institute for Health Information Discharge Abstract Database: a validation study. Accessed January 2, 2025. Available at: https://www.ices.on.ca/publications/atlases-and-reports/2006/canadian-institute-for-health-information
- 12.↑
Government of Canada. Canada Health Act. Accessed February 29, 2020. Available at: https://www.canada.ca/en/health-canada/services/health-care-system/canada-health-care-system-medicare/canada-health-act.html
- 13.↑
Bethell J, Rhodes AE. Identifying deliberate self-harm in emergency department data. Health Rep 2009;20:35–42.
- 14.↑
Mahar AL, Cramm H, Aiken AB, et al. A retrospective cohort study comparing non-fatal self-harm emergency department visits between Canadian veterans living in Ontario and matched civilians. Int Rev Psychiatry 2019;31:25–33.
- 15.↑
Mason SA, Nathens AB, Byrne JP, et al. Association between burn injury and mental illness among burn survivors: a population-based, self-matched, longitudinal cohort study. J Am Coll Surg 2017;225:516–524.
- 17.↑
Karlj B. Measuring “rurality” for purposes of health-care planning: an empirical measure for Ontario. Ont Med Rev 2009:33–52.
- 18.↑
Krieger N. Overcoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology. Am J Public Health 1992;82:703–710.
- 19.↑
Matheson FI, Dunn JR, Smith KLW, et al. Development of the Canadian Marginalization Index: a new tool for the study of inequality. Can J Public Health 2012;103(Suppl 2):S12–16.
- 20.↑
Elixhauser A, Steiner C, Harris DR, Coffey RM. Comorbidity measures for use with administrative data. Med Care 1998;36:8–27.
- 21.↑
Austin SR, Wong YN, Uzzo RG, et al. Why summary comorbidity measures such as the Charlson Comorbidity Index and Elixhauser score work. Med Care 2015;53:e65–72.
- 22.↑
Gutacker N, Bloor K, Cookson R. Comparing the performance of the Charlson/Deyo and Elixhauser comorbidity measures across five European countries and three conditions. Eur J Public Health 2015;25(Suppl 1):15–20.
- 23.↑
Edge SB, Compton CC. The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM. Ann Surg Oncol 2010;17:1471–1474.
- 24.↑
Klaassen Z, Wallis CJ, Chandrasekar T, et al. Cancer diagnosis and risk of suicide after accounting for prediagnosis psychiatric care: a matched‐ cohort study of patients with incident solid‐organ malignancies. Cancer 2019;125:2886–2895.
- 25.↑
Snapinn SM, Jiang Q, Iglewicz B. Illustrating the impact of a time-varying covariate with an extended Kaplan-Meier estimator. Am Stat 2005;59:301–307.
- 26.↑
Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology 1999;10:37–48.
- 27.↑
Wang YH, Li JQ, Shi JF, et al. Depression and anxiety in relation to cancer incidence and mortality: a systematic review and meta-analysis of cohort studies. Mol Psychiatry 2020;25:1487–1499.
- 28.↑
Walker J, Mulick A, Magill N, et al. Major depression and survival in people with cancer. Psychosom Med 2021;83:410–416.
- 29.↑
Nguyen L, Hallet J, Eskander A, et al. The impact of a cancer diagnosis on nonfatal self-injury: a matched cohort study in Ontario. CMAJ Open 2023;11:E291–297.
- 30.↑
Mahar AL, Kurdyak P, Hanna TP, et al. The effect of a severe psychiatric illness on colorectal cancer treatment and survival: a population-based retrospective cohort study. PloS One 2020;15:e0235409.
- 31.↑
Mahar AL, Kurdyak P, Hanna TP, et al. Cancer staging in individuals with a severe psychiatric illness: a cross-sectional study using population-based cancer registry data. BMC Cancer 2020;20:476.
- 32.↑
Chang WH, Lai AG. Cumulative burden of psychiatric disorders and self-harm across 26 adult cancers. Nat Med 2022;28:860–870.
- 33.↑
Barker LC, Sunderji N, Kurdyak P, et al. Urgent outpatient care following mental health ED visits: a population-based study. Psychiatr Serv 2020;71:616–619.
- 34.↑
Hallet J, Sutradhar R, Isenberg-Grzeda E, et al. Association of patient- reported outcomes with subsequent nonfatal self-injury after a new cancer diagnosis. JAMA Oncol 2022;8:e220203.
- 35.↑
Pereira JL, Chasen MR, Molloy S, et al. Cancer care professionals’ attitudes toward systematic standardized symptom assessment and the Edmonton Symptom Assessment System after large-scale population-based implementation in Ontario, Canada. J Pain Symptom Manage 2016;51:662–672.e8.
- 36.↑
Snyder C, Wu AW, eds. Users’ Guide to Integrating Patient-Reported Outcomes in Electronic Health Records. Johns Hopkins University; 2017. Accessed September 7, 2022. Available at: https://www.pcori.org/sites/default/files/PCORI-JHU-Users-Guide-To-Integrating-Patient-Reported-Outcomes-in-Electronic-Health-Records.pdf
- 37.↑
Aaronson N, Elliott T, Greenhalgh J, et al. User’s guide to implementing patient-reported outcomes assessment in clinical practice. Accessed September 7, 2022. Available at: https://www.isoqol.org/wp-content/uploads/2019/09/2015UsersGuide-Version2.pdf
- 38.↑
Montgomery N, Howell D, Ismail Z, et al. Selecting, implementing and evaluating patient-reported outcome measures for routine clinical use in cancer: the Cancer Care Ontario approach. J Patient Rep Outcomes 2020;4:101.
- 39.↑
Di Maio M, Basch E, Denis F, et al. The role of patient-reported outcome measures in the continuum of cancer clinical care: ESMO clinical practice guideline. Ann Oncol 2022;33:878–892.
