OBJECTIVES: There has been a significant amount of research performed on the relationship between the presence of chronic pain and all forms of suicidality. This study explored which rehabilitation acute pain patient (APP) and rehabilitation chronic pain patient (CPP) variables are predictive of six suicidality items: wanting to die because of pain; wanting to die because life is hard; history of wanting to die; history of suicide attempts; recent frequent suicide ideation; and having a suicidal plan. METHODS: The six suicide items were contained within the Battery for Health Improvement-Research Version (BHI-R) and were administered to a healthy community sample (n = 1478), community patients (n = 158), rehabilitation APPs (n = 326), rehabilitation CPPs (n = 341) and rehabilitation patients without pain (n = 110). RESULTS: Affirmation of the six items in APPs/CPPs ranged from 6.13% to 34.90%. Logistic regression predictor models for each item for APPs/CPPs were developed utilizing available variables from the BHI-R. Predictor variables differed between APPs and CPPs and between items. Most predictor variables had previously been delineated in non-pain behaviour studies (e.g., substance abuse). Some novel variables such as perseverance were also identified. Contrary to the behaviour suicide literature, no demographic variables (except employment) were predictive. Correct patient classification ranged from 87% to 95%, in most cases being better than the base rate prediction. DISCUSSION: Suicidality predictor variables were differentially distributed between APPs and CPPs and between different forms of suicidality. Some suicidality predictor variables appeared to be specific to pain patients.
OBJECTIVES: There has been a significant amount of research performed on the relationship between the presence of chronic pain and all forms of suicidality. This study explored which rehabilitation acute painpatient (APP) and rehabilitation chronic painpatient (CPP) variables are predictive of six suicidality items: wanting to die because of pain; wanting to die because life is hard; history of wanting to die; history of suicide attempts; recent frequent suicide ideation; and having a suicidal plan. METHODS: The six suicide items were contained within the Battery for Health Improvement-Research Version (BHI-R) and were administered to a healthy community sample (n = 1478), community patients (n = 158), rehabilitation APPs (n = 326), rehabilitation CPPs (n = 341) and rehabilitation patients without pain (n = 110). RESULTS: Affirmation of the six items in APPs/CPPs ranged from 6.13% to 34.90%. Logistic regression predictor models for each item for APPs/CPPs were developed utilizing available variables from the BHI-R. Predictor variables differed between APPs and CPPs and between items. Most predictor variables had previously been delineated in non-pain behaviour studies (e.g., substance abuse). Some novel variables such as perseverance were also identified. Contrary to the behaviour suicide literature, no demographic variables (except employment) were predictive. Correct patient classification ranged from 87% to 95%, in most cases being better than the base rate prediction. DISCUSSION: Suicidality predictor variables were differentially distributed between APPs and CPPs and between different forms of suicidality. Some suicidality predictor variables appeared to be specific to painpatients.