Association between Adverse Social Determinants of Health and Suicide Death

Introduction

Suicide is a significant public health problem. In the United States, suicide is one of the leading causes of death among individuals aged 10 to 64 years. In 2020 alone, nearly 46,000 individuals died by suicide, equating to one death every 11 minutes.¹ Suicide can be prevented by addressing risk factors among high-risk individuals. As supported by the Social-Ecological Model,² risk factors for suicide exist at multiple levels: individual (e.g., physical and mental health disorder diagnoses, legal/financial problems, history of violence or trauma), relationship (e.g., social isolation, high conflict or violence), community (e.g., discrimination, limited healthcare access), and societal (e.g., stigma related to asking for help or seeking treatment for mental health disorders). Efforts to identify individuals at high-risk for suicide must consider the interconnection of these diverse, multi-level factors. Although physical and mental health-related risk factors are readily identifiable in a healthcare setting, these risk factors may not be enough to effectively identify all people at high suicide risk. For example, only about half of individuals who die by suicide have a diagnosed mental health disorder.³ Identification of individuals experiencing risk factors related to social determinants of health (SDoH), defined as the “the nonmedical factors that influence health outcomes […or] the conditions in which people are born, grow, work, live, and age, and the wider set of forces and systems shaping the conditions of daily life,”⁴ may help to identify individuals at high risk for suicide who might be missed using clinical risk factors only. A standardized way to identify individuals experiencing SDoH risk factors at a population-level is necessary for the development of effective efforts to identify individuals at high risk for suicide. Individuals identified as experiencing adverse SDoH could also be targeted by future interventions addressing modifiable SDoH.

Multiple methods could be used to identify SDoH in healthcare settings. Numerous health systems have begun to collect information about SDoH through patient portals or screeners utilized during healthcare visits. Diagnosis codes from the International Statistical Classification of Diseases (ICD) are another standardized way to document SDoH in a healthcare setting. Since diagnosis codes are routinely collected as part of clinical care, identification of individuals experiencing SDoH-related risk factors through diagnosis codes could be easily integrated into existing clinical practices and data. The International Statistical Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) contained a set of V-codes that documented SDoH that might influence receipt of healthcare.

The main purposes of this study were to identify the prevalence of ICD-9-CM codes related to adverse SDoH among individuals who died by suicide between 2000 and 2015 and to examine the associations between documented adverse SDoH and suicide. Although we anticipated that the prevalence of adverse SDoH-related diagnosis codes would be low, given that these codes may have been underutilized during the ICD-9 era,^5–7 we hypothesized that, as supported by past literature,^7–13 adverse SDoH would be associated with increased suicide risk. Observing such associations would provide additional support for increased standardized documentation of SDoH in healthcare settings, including for use in suicide prevention efforts.

Methods

Results

Discussion

This study sought to investigate the association between adverse SDoH, documented using ICD-9-CM codes, and suicide in a general sample of people receiving care in integrated health systems. Among individuals who died by suicide, 6.6% had any documented adverse SDoH. The prevalence of adverse SDoH investigated in our study were lower than would be anticipated based on the prevalence of these issues in the general United States’ population. For example, 4,272 individuals (1.3%) in our sample had documented problems related to housing and economic circumstances. However, 13.5% of the population in the United States lived in poverty in 2015,¹⁴ and, for approximately one-third of United States households in 2015, housing posed a cost burden, composing more than 30% of their income.¹⁵ While 154 individuals (0.5%) in our sample had documented problems related to employment and unemployment or occupational maladjustment, the unemployment rate in the United States at the end of 2015 was 5.0%.¹⁶ Previous studies have also reported a low prevalence of documented adverse SDoH using ICD-9-CM codes.^5–7

We found that presence of any documented adverse SDoH, as well as documentation of family alcoholism or drug addiction, being a victim or perpetrator of abuse, other primary support group problems, care provider dependency, employment or occupational maladjustment problems, housing or economic problems, legal problems, and other psychosocial problems, was associated with increased odds of suicide. These associations are also consistent with previous research. Among veterans, Alemi et al.⁷ found that presence of an SDoH-related V-code in the electronic medical record was associated with, on average, a 24-fold increase in the odds of suicide or intentional self-injury. However, only about 13% of veterans included in their study had at least one V-code documented in their full electronic health record data available through the Department of Veterans Affairs. Although there are strong associations observed between documented adverse SDoH and suicide, the low prevalence of adverse SDoH reported in medical records demonstrates that much work is needed if these risk factors are to be used in identifying and treating those individuals at high risk for suicide.

Given the impact that SDoH have on diverse health-related outcomes, calls have been made to increase the documentation of SDoH in medical records. In 2014, the Institute of Medicine released guidance on capturing SDoH in electronic health records.^17,18 The 2023 Healthcare Effectiveness Data and Information Set (HEDIS) quality measures will include a social needs screening and intervention measure, which evaluates the proportion of members that are screened in three SDoH categories (food, housing, and transportation) and, following a positive screen, receive an appropriate intervention.¹⁹ As part of a national movement, many health systems have started to collect patient-reported information about SDoH during clinical visits, such as through use of a standardized SDoH screening tool embedded in the electronic health record. Better documentation of the burden of adverse SDoH in populations served by specific healthcare institutions would provide improved information about the SDoH-related problems most impacting each community and would support the allocation of additional resources to address these issues. Increased adverse SDoH documentation would also allow for better understanding of the association with between adverse SDoH and specific health outcomes, which could help to identify high-risk individuals for targeted interventions.

Despite its potential benefits, multiple reasons have been given for the observed low prevalence of SDoH documentation in medical records. First, providers may be unlikely to ask about SDoH in clinical settings, perhaps because resources may not be available to help individuals experiencing adverse SDoH.^5,20 Second, diagnosis codes related to SDoH currently generate no revenue, so there is limited financial incentive to include this information in the medical record.⁵ However, the introduction of the HEDIS measure related to social needs might help to combat this issue. Third, ICD-9-CM V-codes failed to include categories for some important SDoH or contained ambiguous categories that combined multiple SDoH together.⁵ These limitations were improved slightly through development of a more detailed list of SDoH-related codes for ICD-10-CM, known as Z codes, but further expansion of the list of SDoH diagnosis codes is still needed.^21,22 Newer SDoH screening tools being implemented in health systems nationwide may also more detailed information about adverse SDoH. Finally, patients may not want information about adverse SDoH included in their medical records due to perceived stigma.

Multiple methods have been proposed to increase documentation of SDoH in medical settings, including development of standardized SDoH screening tools, consistent financial reimbursement for SDoH-related diagnosis codes, endorsements from policymakers and payers supporting recording of these diagnosis codes, and establishment of guidelines and/or mandates for reporting of SDoH in medical records and billing data.^5,23 To allow for effective documentation of SDoH in medical records, multiple personnel, not just clinicians, must be able to document SDoH. New SDoH screening tools embedded within patient portals often can be completed by the patients themselves. SDoH-related questions should be granular enough to capture data directly usable in referrals and interventions (e.g., asking about food, utility, housing, and transportation insecurities separately rather than just asking about general economic insecurity) and should be sensitive to the context in which they will be implemented. Mechanisms to capture SDoH should also be built into existing processes and structures, such as through utilization of existing functionalities in an electronic health record.²³ Through implementation of some of these methods, SDoH information has been successfully integrated into medical records in several settings.^23–25 At Kaiser Permanente Northwest, for example, individuals experiencing SDoH-related issues were identified at several points during clinical contact, including while checking-in for a visit, during the actual clinical visit, or during targeted intervention efforts for high-risk populations, and then received outreach from trained patient navigators. The patient navigators collected SDoH-related data using a standardized assessment tool and provided information about both health system and community resources that could help to address the patient’s SDoH concerns. Data about the encounters between the patient and the patient navigator regarding adverse SDoH were documented in the electronic health record for future reference, including for use in providing healthcare that considered each individual patient’s unique set of circumstances. Use of ICD-10-CM Z codes to document adverse SDoH in the electronic health record allowed for easy data extraction for clinical treatment, reporting, or research.²⁴ Additional research is needed to establish how to capture SDoH data most effectively in medical records using standardized methodology across diverse healthcare settings.

For suicide prevention efforts specifically, data about adverse SDoH captured in medical records could be used both in efforts to identify and treat individuals at high risk for suicide. SDoH data could be incorporated into existing suicide risk prediction models, such as the MHRN’s suicide risk prediction model,²⁶ providing new data that, in combination with demographic, clinical, and healthcare utilization variables already included in the model, could enhance ability to identify individuals at high risk for suicide. Interventions could also be developed to target modifiable SDoH risk factors to reduce suicide risk, such as ongoing relationship stress with a family member, lack of transportation to attend outpatient mental health visits following discharge from a psychiatric hospitalization, or unemployment that may limit the ability to purchase prescription medications to treat a diagnosed mental health disorder.

This study built upon previous literature by utilizing electronic health record data from a large sample of individuals who died by suicide and their matched controls to investigate the prevalence of adverse SDoH documented among individuals who died by suicide and the associations between adverse SDoH and suicide. However, this study also has limitations. First, as not all individuals were enrolled in a given health plan for the full year prior to the index date, any healthcare received outside of that plan, and any adverse SDoH codes included as part of that healthcare, is not captured in this analysis. The findings of this study may also not be generalizable to individuals being served in settings different from those in which our data was collected. Second, this data represents only a small proportion of suicide deaths that occurred during the study period. Suicides may also be misclassified, perhaps as accidental or undetermined deaths.^27,28 Third, our data included only ICD-9-CM V codes. Additional research investigating ICD-10-CM Z code prevalence and the associations between ICD-10-CM Z codes and suicide is needed. Fourth, our sample contained a small number of individuals with reported adverse SDoH. As a result, some of our estimates, although statistically significant, may be unstable, as indicated by wide confidence intervals, and should be interpreted with caution. We were also able to control for only a small number of potential confounding variables in our analysis and could not adjust for some potentially important variables, such as socioeconomic status, healthcare utilization levels, or race/ethnicity. Race/ethnicity information was not available for individuals in this study. Residual confounding may impact our results. Fifth, there may be greater underreporting of some SDoH categories than others in our data if specific ICD-9-CM V codes were more likely to be captured in medical records than others regardless of their actual population prevalence. For example, if issues related to abuse were viewed as more relevant to treatment for mental health disorders than other adverse SDoH categories, corresponding ICD-9-CM V codes may be more likely to be included in claims for those services. Finally, within our data, ICD-9-CM V codes may be recorded more frequently for individuals with the most severe or persistent difficulties, who are also likely at increased suicide risk compared to other individuals served at the health systems included in this study.

Conclusion

Our study identified a low prevalence of documented adverse SDoH among individuals who died by suicide. However, we found that several adverse SDoH were associated with increased odds of suicide, including primary support group problems, being a victim or perpetrator of abuse, or legal, economic, or employment-related problems. Given the strong associations between SDoH and a diverse range of health-related outcomes, increased documentation of SDoH in medical records has been championed recently. Increased documentation of SDoH will improve understanding of the prevalence of adverse SDoH in different populations and will assist in identifying and intervening among individuals at high risk for negative health-related outcomes, including suicide.