Background:To study the variations of the effect of risk factors for suicidal ideation (SI) and suicide attempt (SA) by geographical regions among commercially insured youth and young adults in the US.Methods:A national level retrospective cohort study was conducted using health insurance claims data from four major insurance companies in the US. The cohort was defined as patients having a mental health (MH) or substance use disorder related outpatient encounter (index event). We used survival analysis to evaluate the geographical variation of the effect of risk factors on patients’ future SI and SA after the index event. Risk factors considered in the models consist of a mix of long-, mid-, and short-term prior comorbidities (i.e., depression, drug abuse) and/or prescriptions (i.e., Benzodiazepine, antidepressant) that are identified as important factors for predicting SA by Simon et al. 2018. Patients’ geographical regions were assigned to one of the nine divisions defined by the US Census Bureau: East North Central (Illinois, Indiana, Michigan, Ohio, and Wisconsin), East South Central (Alabama, Kentucky, Mississippi, and Tennessee), Mid-Atlantic (New Jersey, New York, and Pennsylvania), Mountain (Arizona, Colorado, Idaho, Montana, Nevada, New Mexico, Utah, and Wyoming), New England (Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont), Pacific (Alaska, California, Hawaii, Oregon, and Washington), South Atlantic (Delaware, Florida, Georgia, Maryland, North Carolina, South Carolina, Virginia, District of Columbia, and West Virginia), West North Central (Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, and South Dakota), and West South Central (Arkansas, Louisiana, Oklahoma, and Texas). Patients were censored if they did not have an SI or SA diagnosis before 9/30/2015 (ICD-9 to ICD-10 switch date) or the last day of the insurance plan enrollment, whichever came first.Results:Between 2014-2015, we identified 317,383 commercially insured patients with mental health coverage of age <65 years old with an index encounter. The prevalence of SI was the highest in the Mountain region (6.98%; N=1657) and the lowest in the East South-Central region (3.56%; N=608). The prevalence of SA was the highest in the Mountain region (1.89%; N=448) and the lowest in the Mid-Atlantic region (0.87%; N=398). Results of Cox proportional hazards models showed significant geographic variations of the effect of risk factors on both SI and SA (p<0.001 for both SI and SA).For SI, there was regional variation of effects of rurality, alcohol use disorder diagnosis in the past 3 years, eating disorder diagnosis in the past 3 years, anxiety disorder diagnosis in the past 3 years, and MH inpatient stay in the past year. For individuals from the Mountain and West South-Central regions, living in rural areas was a protective factor of having SI (HRs=0.72 and 0.66, respectively). For individuals from the Mid-Atlantic and Mountain regions, having an eating disorder diagnosis in the past 3 years was a protective factor of having SI (HRs=0.70 and 0.74, respectively). Having anxiety disorder diagnosis in the past 3 years was positively associated with having SI, except for patients in East South Central and North West Central regions.For SA, there was regional variation of effects of age and having SA in the past year. In general, younger patients were more likely to have SA as compared to older patients. The effects of SA in the past year on future SA varied by geographical region, however none of them were statistically significant.Discussion:Among commercially insured patients in the US, residents of the Mountain region had the highest prevalence of both SI and SA. The effects of important risk factors on SI and SA varied by geographical region.
Read full abstract