Screening for the General Primary Care Population

The Symptom Driven Diagnostic System for Primary Care (SDDS-PC),⁴⁸ a 62-item self-report instrument designed to help identify psychiatric illness in primary care settings, contains 3 items assessing suicide risk.²³ In this sample of patients between 18 and 70 years of age, 2.4% (67/2,749) of patients reported “feeling suicidal” within the past month. At 1 of the 3 sites (n = 1,001 participants), data on suicidal thoughts, plans, and past attempts were systematically collected using a nurse-administered, face-to-face structured interview conducted immediately before the medical visit. The individual operating characteristics of 3 items (“thoughts of death,” “wishing you were dead,” and “feeling suicidal,” within the past month) were determined compared to a structured interview for identifying a plan to commit suicide (the gold standard). “Thoughts of death” had 100% sensitivity, 81% specificity, and 5.9% positive predictive value for detecting patients with a plan to commit suicide. Endorsing “wishing you were dead” had 92% sensitivity, 93% specificity, and 14% positive predictive value; “feeling suicidal” had 83% sensitivity, 98% specificity, and 30% positive predictive value for identifying patients with a plan to kill themselves. Of those “feeling suicidal,” 85% had a psychiatric disorder as determined by structured clinical interviews,⁴⁹ modified to Diagnostic and Statistical Manual of Mental Disorder - Fourth Edition (DSM-IV) criteria.⁵⁰ After adjusting for potentially confounding effects of psychiatric comorbidity and demographics, only major depression (odds ratio [OR] = 33.1; 95% confidence interval [CI], 10.9–99.6) and drug abuse or dependence (OR = 16.7; 95% CI, 3.9–71.4) were independently associated with suicidal ideation. Of note, only 46% of those eligible for this study agreed to participate.

Two additional assessment measures have been used in primary care settings; they appear to be particularly reasonable considerations for testing in future research.^{22, 51} Neither instrument met our inclusion criteria because no research has assessed their test characteristics in a primary care setting. The Scale for Suicide Ideation (SSI), a widely used 21-item, interviewer-administered rating scale that measures the current intensity of patients' attitudes, behaviors, and plans to commit suicide on the day of the interview, is 1 of only 2 suicide risk tools with documented predictive validity for adults seeking outpatient psychiatric treatment.⁵¹ The test takes approximately 10 minutes to administer. Patients who scored in the higher risk category (total score > 2) were approximately 7 times more likely to commit suicide than those who scored in the lower risk category.⁵¹ It has been used as a measure of suicide risk in a variety of medical settings (including the primary care setting), but its sensitivity and specificity have not been assessed in the primary care setting. A related measure with documented predictive validity in adult psychiatric outpatients, the Scale for Suicidal Ideation-Worst (SSI-W), indicated that patients scoring in the higher risk category were 14 times more likely to commit suicide than patients in the lower risk category.⁵

Cooper-Patrick et al. used stepwise logistic regression to create retrospectively a 4-item, interviewer-administered screen, the Suicidal Ideation Screening Questionnaire, using data from the NIMH Epidemiologic Catchment Area study.²² The authors derived this tool from a subgroup of patients who had received care in the general medical sector within the past 6 months. Assessing sleep disturbance, mood disturbance, guilt, and hopelessness correctly identified 84% of patients who endorsed suicidal ideation within the prior 12 months. However, the items do not assess any suicide-related behaviors, the screen is not useful for risk prediction because it does not allow determination of the chronology of events (i.e., whether the four items precede suicidality, follow it, or occur concurrently), and the screen's test characteristics have not been prospectively studied in a primary care setting.

Screening for the General Primary Care Population: Children and Young Adolescents

Despite a plethora of instruments to assess suicide risk specifically among children and adolescents,⁴⁶ we did not find any appropriately evaluated screening tools to assess for risk of suicide among adolescents in general primary care clinic populations.

One recent report by Horowitz et al. described the development of a 4-item screening instrument to identify adolescents at risk of suicide in emergency room settings, a frequent de facto primary care setting.⁴⁷ The items were “Are you here because you tried to hurt yourself?” “In the past week, have you been having thoughts about killing yourself?” “Have you ever tried to hurt yourself in the past, other than this time?” and “Has something very stressful happened to you in the past few weeks?” These 4 items were evaluated against the Suicide Ideation Questionnaire (SIQ) and a shortened version for youth in less than the tenth grade (SIQ-JR) as the criterion standard.⁵² The majority of youth (75%) in this study were between 11 and 16 years of age. This instrument was found to have a sensitivity of 98%, specificity of 37%, positive predictive value of 55%, and negative predictive value of 97%. Of note, subjects in this study were not an unselected primary care population but rather presented with a psychiatric chief complaint, a fact that makes these results less generalizable to routine screening in unselected primary care or emergency room populations. How well this screening instrument performs in general clinic settings has not been tested, and items may need to be modified for use in primary care.

Screening in a High-Risk Primary Care Population

No instruments have been assessed that screen for suicide risk in high-risk primary care populations. These would include persons with a history of self-harm, those with depressive illness, or those with substance abuse. It is this high-risk group that has received the greatest amount of intervention study (see Key Questions 4 and 5 below).

Challenges of Finding a Clinically Useful Screening Measure

Given the rarity of suicide attempts in the primary care population, finding an accurate screening strategy for suicide risk for the general population in a primary care setting is a daunting challenge. This is illustrated, for example, by the following hypothetical situation. Consider a screening instrument (for example, endorsing “feeling suicidal”) that identifies patients at high risk and has reasonable test characteristics (e.g., sensitivity of 80% and specificity of 70%, figures similar to screens for depression.^{53, 54} Apply this tool to a population of 10,000 in which 10 patients will attempt suicide (10-fold more than the 10 in 100,000 persons who will complete suicide). In such a hypothetical population, use of this suicide screening tool to identify those who will make a suicide attempt will produce 8 true positives, 2 false negatives, and 2,997 false positives, a positive predictive value of 0.3%. The high proportion of false positives can generate a substantial burden. Notable disadvantages to such a test would include the patient and provider time needed to follow-up such a positive screen with a confirmatory procedure in primary care practices already pressed for time, the absence of a clear evidence-based decision to guide the follow-up for such a positive screen in primary care, and the financial costs of unnecessary referrals to a mental health provider.

Summary

Only 1 screening instrument to identify those at risk of suicide in primary care settings has been tested, with 1 item (“feeling suicidal”)from the SDDS-PC instrument⁴⁸ corresponding closely with plans to attempt suicide.²³ This instrument was tested on patients from 18 to 70 years of age. Although this result is promising, further study is required before any recommendations can be made regarding the use of screening instruments to identify suicide risk in primary care settings. In children or young adolescents, no testing of instruments in primary care settings has yet taken place.

Randomized Controlled Trials

Thirty-three randomized controlled trials (RCTs) met our inclusion criteria; all involved high-risk groups. Thirty-one of these trials required recent deliberate self-harm (DSH). Two trials (one of which produced 2 articles)^{55, 56} did not require DSH but enrolled patients with borderline personality disorder, a group at increased risk of DSH and of whom at least 75% in each study had a history of DSH.^55–57 One study recruited some of its patients from a primary care setting, although the intervention was performed in a psychiatric outpatient setting.⁵⁷ One study had the intervention performed within a primary care setting.⁵⁸

Trials focusing on the 2 populations of greatest clinical concern (i.e., adolescents/young adults and elderly adults) were limited. We review the three studies that exclusively involved adolescents 17 years of age and younger⁴² as a group.^{43, 44} Of these, 2^{42, 43} had been included in the Hawton et al. review,¹¹ and 1 was published subsequently.⁴⁴ The remaining 30 included either adults only or adults and older adolescents but did not differentiate further by age in the analyses; we review them together.

We found no published intervention study for the geriatric population. However, the Prevention of Suicide in Primary Care Elderly - Collaborative Trial (PROSPECT) is currently being conducted.⁵⁹ This trial aims to determine whether placement of a depression health specialist in primary care practices will have a favorable impact on rates of depression, hopelessness, and suicidal ideation in elderly primary care patients with major or persistent minor depression. Both identification of suicidal ideation and management of suicide risk occur in the primary care setting.⁶⁰ Initial outcomes for the 4- and 8-month follow-up periods are expected by early 2003.

We organize our review of the 33 RCTs as follows. First, we summarize the results of the 21 studies of DSH systematically reviewed by Hawton et al. that involved older adolescents and adults¹¹ (Table 4, see page 41). Of these, 5 dealt with problem-solving therapies compared to standard aftercare, 5 with intensive care plus outreach compared to standard aftercare, 1 each with emergency care or dialectical behavioral therapy compared to standard aftercare, 1 each with 4 different nonpharmaceutical therapies, and 5 with pharmaceutical interventions (4 antipressants, 1 antipsychotic).

We then provide greater detail on the 9 additional RCTs of DSH involving older adolescents and adults that we identified in our literature search^{55–58, 61–66} (Table 5, see page 42). Lastly, we review the 3 studies exclusively involving adolescents 17 years of age and younger (Table 6, see page 43).

Table

Table 5. Additional Randomized Controlled Trials of Interventions to Reduce Deliberate Self-Harm in Adults and Older Adolescents*.

Table

Table 6. Randomized Controlled Trials to Decrease Deliberate Self-Harm in Young Adolescents Exclusively.

Older Adolescents and Adults: Prior Review of Randomized Clinical Trials for Deliberate Self-Harm

Twenty-one studies of adults receiving treatments for deliberate self-harm are described below (Table 4);^{13, 67–86} of these, 12 included older adolescents.^{13, 67–74, 76, 81, 83} Although some trends suggested incremental benefit from certain interventions, compared to usual care, interventions for which more than 1 study was performed produced no statistically significant effects by meta-analysis. The most promising intervention was problem-solving therapy, a short-term cognitively-oriented psychotherapy, which was tested in patients 15 years of age and older. For these studies, the summary odds ratio (OR) showed a trend toward decreasing DSH for 5 studies of problem-solving therapy versus standard aftercare (OR 0.70; 95% CI, 0.45–1.11).^67–71

Intensive care plus outreach versus standard aftercare (6 studies)^{13, 72–76} produced a summary OR of 0.83 (0.47–1.48).

One large trial of emergency care consisting of providing the patient with a card with physician contact information and an offer of crisis intervention as needed was compared to standard care.⁷⁷ Results showed a trend toward a decreased likelihood of repeating DSH in favor of the intervention (0.43, 95% CI 0.15–1.27).

Two interventions, dialectical behavior therapy and flupenthixol (an antipsychotic), reported statistically significant reduced repetition of DSH relative to standard care or placebo, respectively, but, involved a maximum of 20 patients in each group. Dialectical behavior therapy (DBT), a comprehensive treatment program developed to treat chronic and severely dysfunctional individuals with borderline personality disorders by improving emotional and behavioral management skills, significantly reduced repetition of DSH for patients ages 18 to 45 years with borderline personality disorder and recent DSH relative to standard care (OR = 0.24; 95% CI, 0.06–0.93).⁷⁸ Administration of the antipsychotic flupenthixol significantly reduced the proportion of repeated DSH for those ages 18 to 68 years with a history of at least 2 prior suicide attempts as compared with placebo (OR = 0.09; 95% CI, 0.02–0.50).⁸²

Older Adolescents and Adults: Additional Randomized Controlled Trials of Deliberate Self-Harm

Nine additional studies involving repetition of DSH in adult populations were identified in our literature search (Table 5).^{55–58, 61–66} One study showed significant benefit from interpersonal psychotherapy (IPT), a time-limited psychotherapy that focuses on resolving current interpersonal problems to improve symptoms, relative to standard care. In this study, Guthrie et al. recruited patients ages 18 to 65 years from an emergency room setting who presented with deliberate self-poisoning but did not require medical or psychiatric hospitalization.⁶¹ Patients were randomized to either 4 50-minute sessions of IPT delivered by nurse therapists in the patient's home or to usual care. Fifty-one percent (n = 119) of those eligible participated, with those refusing being at a greater suicide risk of suicide as indicated by their being more likely to have a history of self-harm, to have left a suicide note, and to express the wish to die. In an intention-to-treat analysis, those in the IPT group were less likely to have a repeat episode of DSH in the subsequent 6-month period (8.6% v. 27.9%, P = 0.0009).

Bateman et al. compared psychoanalytically oriented partial hospitalization to standard aftercare for patients ages 16 to 65 years with borderline personality disorder who were followed in psychiatric outpatient clinics.^{55, 56} Treatment occurred for a maximum of 18 months. Although inclusion criteria did not require DSH, ninety-five percent of those randomized to partial hospitalization and at least three-quarters of those randomized to usual care had a history of prior suicide attempts, indicating that both groups were at high risk of DSH.

Analysis was not intention-to-treat. Twenty-two patients were initially randomized to each group. Three patients in the partial hospitalization group were lost to follow-up, but 3 patients were allowed to switch from the usual care group to partial hospitalization, producing 22 patients in the partial hospitalization group and 19 in the usual care group. After 18 months of treatment, the percent of those with suicide attempts within the prior 6 months was significantly lower in the treatment group than the control group (53% for intervention group, no rate given for control group but figure suggests approximately 40%, P < 0.001 by Mann-Whitney test). At 18-month follow-up after initial treatment regimen, a significantly smaller proportion of the partial hospitalization group made a suicide attempt (4/22, or 18%) than did the usual care group (12/19, or 63%; no OR given, P < 0.004 by Fisher's exact test). If only the original 19 randomized to partial hospitalization were included in the analysis, the difference remained significant (P < 0.001 by Mann-Whitney test).

The remaining 7 studies identified no benefit from interventions beyond either standard care^{57, 58, 62, 64, 65}or placebo.^{63, 66}

Evans M. et al. enrolled patients after discharge from medical hospitals who had been admitted following DSH and referred for psychiatric consultation.⁶⁴ Patients were randomized to receive either an emergency information card providing the number for 24-hour crisis telephone consultation with an on-call psychiatrist for 6 months (n = 417) or usual care (n = 410). Outcome information was determined by subsequent visits to one of the three participating hospitals, at which time they were included in a computerized case registry. This system detected overdose relatively well but was less reliable in detecting self-laceration. In an intention-to-treat analysis at 6-month follow-up, the groups did not differ significantly in the likelihood of repeating an attempt (OR = 1.20; 95% CI, 0.82–1.75), nor did they differ in the number of repeated episodes. In a subgroup analysis of patients dichotomized by prior history of DSH, those with a previous history of DSH in the intervention group had higher odds of repeating (OR = 1.85; 95% CI, 1.14–3.03). Those with no prior history experienced a nonsignificant protective effect (OR = 0.64; 95% CI, 0.34–1.22).

Motto and Bostrom tested a low-intensity outpatient intervention (brief contact by letter) to usual care in a group of patients who had been admitted to an inpatient psychiatric facility either depressed or suicidal and who had declined therapy after hospital discharge.⁶⁵ The intervention group received a brief contact letter once per month for 4 months, followed by once every 2 months for 8 months, and then once every 3 months for 4 years. The control group received no letter. The outcome of interest was suicide, as measured by health statistics, clinical sources, family members, death certificates, and coroners' records; analysis was intention-to-treat. At 5-year follow-up, intervention and control groups did not significantly differ in the proportion of patients who completed suicide (3.9% v. 4.6%). Contact during the first 2 years of the study was found by Kaplan-Meier survival analysis to decrease significantly the risk of suicide attempts over that time period, but no benefit accrued by the 5-year follow-up.

Rudd et al. compared an outpatient day hospitalization program to usual outpatient care for a group of military patients recruited from inpatient and outpatient military settings.⁶² Eligible patients were either referred following a suicide attempt, had a mood disorder with current suicidal ideation, or had episodic alcohol abuse with current suicidal ideation. Intervention occurred for 2 weeks and follow-up occurred at 1 year. The attrition rates were high (75% in experimental group and 79% in control group), and 20% of the intervention group was hospitalized. Analysis was not intention-to-treat. The authors compared multiple measures of suicidal ideation and behavior and found no differences between the 2 groups; both groups improved over time.

Montgomery et al. compared antidepressants to placebo for psychiatric outpatients with a history of at least 2 prior suicide attempts but no evidence of major depression.⁶³ Patients were seen by psychiatrists twice a week for 6 months and randomized to either fluoxetine treatment at 60 mg twice a week (120 mg/week total, n = 54) or placebo (n = 53). Analysis was not intention-to-treat. The percentage of patients with repeat attempts at 6 months did not differ significantly (33.3% v. 34%). This dosing is substantially below the standard dosing of 20 mg per day for patients with depressive illness.

Battaglia et al. tested the benefit of low-dose intramuscular fluphenazine (an antipsychotic) compared to placebo (ultra-low dose intramuscular fluphenazine) in a group of patients evaluated in a psychiatric emergency room who had made a suicide attempt within the prior 30 days and who had a history of at least 2 prior suicide attempts.⁶⁶ Patients receiving or expected to receive any other psychotropic medications were excluded from the study. The outcome measured was the change in the rate of serious DSH per month over the prior 6 months. The 2 groups did not differ significantly (-0.16 v. -0.06, P = 0.1459 by Mann-Whitney test). Analysis was not intention-to-treat.

Koons et al. compared dialectical behavioral therapy (DBT) to usual care in a six-month RCT involving women veterans with borderline personality disorder.⁵⁷ Patients were recruited from one Women Veterans primary care clinic, multiple Veterans Counseling Centers, and other Veteran Medical Centers within the same state. A total of twenty-eight women were randomized; 10 within each arm completed the study and were analyzed. The rate of lifetime history of DSH among all enrolled was reported to be 75%, with no significant difference between the two groups studied. Six months after beginning treatment, there was no significant difference between the two groups in the proportion with DSH in the previous 3-months; 1/10 women in the DBT arm had an episode of DSH in the previous 3-months (relative to a baseline rate of 50%) compared to 2/10 in the control arm (relative to a baseline rate of 30%). Of note, the usual care group, with unfettered access to weekly individual therapy and supportive and psychoeducational groups, received an intensity and quality of care that is likely greater than most usual care groups.

In the only study to test an intervention for suicide risk in the primary care setting, Bennewith et al. compared a 3-part, 1 time intervention to usual care.⁵⁸ Providing the intervention involved general practitioners with (1) a letter informing them of a patient's DSH episode, (2) a letter the physicians at their discretion could forward to the patient inviting him or her to make an appointment, and (3) guidelines on the assessment and management of DSH in general practice, to usual care. Eligible patients were those with a new episode of DSH (identified from a case register in a weekly report from a local hospital's accident and emergency department) who were already patients within the participating practices. In an intention-to-treat analysis at 12-month follow-up, the groups did not differ significantly in the proportion of patients who attempted suicide (21.9% v 19.5%). Of note, adherence to this low-intensity intervention was poor; only 58% of the intervention group physicians sent a letter to the patient within a 12-month period.

Children or Adolescents 17 Years of Age or Younger Exclusively: Randomized Controlled Trials of Deliberate Self Harm

We identified 3 RCTs conducted exclusively in patients 17 years of age or younger (Table 6),^42–44 2 of which^{42, 43} were previously reviewed by Hawton.¹¹ Harrington et al. compared a home-based family intervention plus usual care versus usual care alone for treatment of adolescents age 16 years or younger referred to psychiatric hospitals in the United Kingdom for deliberate self-poisoning.⁴³ The home-based family intervention involved 1 assessment session plus 4 home visits; follow-up lasted for 6 months. The intervention did not influence the repetition of DSH (OR, 1.02; 95% CI, 0.41–2.51). Analysis was not intention-to-treat.

Cotgrove et al. studied the benefit of using an emergency card in a population of 105 children between the ages of 12.2 and 16.7 years for a period of 12 months.⁴² Patients were recruited from inpatient hospitals following a DSH episode. In an intention-to-treat analysis, those randomized to the emergency card group (the card acted as a “passport” to readmission to a pediatric hospital ward) showed a tendency toward decreased DSH (OR, 0.50; 95% CI, 0.12–2.04), an estimate that is similar to that from the Morgan et al. adult study of emergency cards (Table 4).⁷⁷ Both effects were in the opposite direction from the largest adult study of the use of emergency cards involving 817 adults by Evans M. et al., which reported a point estimate of 1.20 (95% CI 0.82–1.75) (Table 5).⁶⁴

Wood et al. assessed group therapy versus standard care in a population of 12-to-16-year-old outpatients referred for mental health services as a result of DSH.⁴⁴ The experimental intervention consisted of an initial assessment phase of 6 group sessions followed by weekly group sessions until the patient felt ready to leave. Individual sessions were available if needed. In an intention-to-treat analysis 7 months after randomization, those in group therapy (n = 32) were significantly less likely than those in usual care (n = 31) to be “repeaters” (defined as having 2 or more episodes of DSH post-randomization) (6% v. 32%; OR, 0.16; 95% CI, 0.03–0.71). Adolescents randomized to group therapy tended to have fewer episodes of DSH than those allocated to routine care (mean 0.6 v. 1.8, statistical analysis not reported), but this finding did not reach statistical significance. Despite reliance on self-report as the main measure of DSH, the strength of these findings within the context of a small sample size is promising.

Cohort Studies

Of the 4 cohort studies that met our inclusion criteria, 2 were in adults and older adolescents,^{87, 88} and 2 were in child or adolescent populations exclusively^{89, 90} (Table 7, see page 44). None of these studies produced statistically significant differences involving repeated suicidal behavior.

Table

Table 7. Cohort Studies to Decrease Suicidal Behavior in At-risk Patients*.

Older Adolescents and Adults: Cohort Studies

Coryell et al. evaluated suicide risk in a long-term cohort of patients with major affective disorders diagnosed at 1 of 5 academic sites in the United States;⁸⁷ the mean age was approximately 40 years, and subjects were 17 years or older. From this cohort, a nested case-control study was performed in which the first case group consisted of 15 patients who committed suicide and the second case group of 41 patients who made a serious suicide attempt. The controls were nonsuicidal patients who were also from the cohort. All cases were receiving some type of therapy at the time of the suicide or suicide attempt. The controls were matched to the cases on sex, polarity (i.e., whether depressed or manic) at the time of suicide or suicide attempt, lifetime diagnoses of substance abuse or alcoholism, history of a prior serious suicide attempt, and receipt of a similar composite antidepressant score (measured by similarity in strengths and types of antidepressants) to the case. Cases were compared to controls on use of lithium in the week before the suicide or suicide attempt, using the case's exposure time to designate the exposure time in the controls. The investigators found no relationship between use of lithium before the suicide or suicide attempt. The cases appeared to have more severe disease based on the duration of cohort enrollment, percentage with a history of prior suicide attempts, and use and type of antidepressants at the time of assessment.

Raj et al. compared the use of 10 sessions of cognitive-behavioral counseling over a 2–3 month period to routine medical treatment for patients admitted to the intensive care unit (ICU) of a general hospital following their first or second suicide attempt.⁸⁸ The patients were between 16 and 50 years of age and needed to have coexisting depression or anxiety. Upon admission to the ICU, the 40 enrolled patients were sequentially assigned to either the counseling intervention or to routine medical care with the option to attend therapy sessions. None of the intervention group repeated a suicide attempt at 2 to 3 months of follow-up; 1 patient in the control group made a repeat suicide attempt. The significance of this small absolute difference is unclear.

Children and Adolescents 18 years of Age or Younger Exclusively: Cohort Studies

Pfeffer et al. followed a cohort of 69 children and adolescent psychiatric inpatients receiving routine aftercare (53 of whom reported suicidal ideation or an attempt within the 6 months prior to hospitalization) and a group of 64 community subjects (who were not psychiatric patients and were matched on demographic characteristics) for 6 to 8 years.⁸⁹ At initial assessment the cohort was between 4.7 and 14.7 years of age (mean 10.5 +/- 1.8 years). Information about type, onset, and duration of psychiatric treatment was obtained from multiple sources, including subjects, parents, and records of medical facilities and schools. In the follow-up period, 20 suicide attempts and no deaths occurred; no direct comparison of two groups by suicide attempts was reported. The investigators found no association between annual rates of combined psychiatric services and the occurrence of a suicide attempt post-hospitalization or between type of services received and time to first suicide attempt post-hospitalization. They did find a positive association between use of antidepressant medication and time to occurrence of a suicide attempt in the follow-up period, which the authors speculate is related to the fact that more severely ill patients are preferentially prescribed medication.

Rotheram-Borus et al. evaluated an emergency room intervention targeting both urban Latino females, ages 12 to 18 years, who presented with a suicide attempt, and their mothers.⁹⁰ The 3-component intervention involved emergency room staff training, adolescent-mother pairs viewing a video, and a family therapy session before leaving the emergency room. During the 18-month follow-up period, 6 repeat suicide attempts occurred in the intervention group and 11 in the standard treatment group, a difference that did not reach statistical significance. The intervention was associated with less depression among adolescents and increased use of follow-up outpatient therapy.

Summary

Among these studies, we saw no statistically significant effects for interventions for which more than 1 study of the intervention had been performed. Some trends, however, suggest incremental benefit from some interventions (in particular, problem-solving therapy for patients 15 years of age and older). Of the interventions for which only 1 study was done, the most promising are DBT for borderline personality disorder in adults, interpersonal psychotherapy (IPT) for DSH in adults (ages 18 to 45 years), and group therapy for DSH in younger adolescents (ages 12 to 16 years). These results, however, need further confirmation.

The evidence basis has 3 primary limitations. First, the studies tend to be underpowered, so that there is risk of falsely concluding that a particular intervention does not produce a statistically significant benefit when in fact such a benefit exists. This limitation is understandably a consistent problem given the rarity of the event.

Second, standard care, the most common comparison group used in the studies, is poorly described. It likely varies across the multiple studies, making it unclear to what the experimental intervention is really being compared. The components of standard care are poorly understood; indeed, standard care itself in many cases may be an effective intervention.

Third, inconsistent age ranges, and lack of stratification based on age, limits our ability to make meaningful conclusions specific to particular age groups. This limitation may reflect the fact that one of the highest risk age ranges, persons 15 to 24 years, includes both older adolescents and young adults. Stratification within this age group in subsequent studies could better address this question.

Older Adolescents and Adults: Prior Review of Randomized Controlled Trials Involving Intermediate Outcomes

Townsend et al. conducted a systematic review of 6 RCTs^{67–71, 93} involving brief problem-solving therapy in patients with DSH in which the outcomes included depressive severity, hopelessness, and improvement in problems (Table 8).⁴¹ The analyses did not stratify the study populations by age. The 4 studies that evaluated depressive outcomes^{67–69, 71} used 2 different scales for depression, requiring Townsend et al. to calculate a Standardized Mean Difference (SMD; the mean difference divided by the pooled sample standard deviation, producing results related to multiples of the standard deviation) to evaluate depressive symptoms. The summary SMD indicated a significantly lower depression score of about one-third of a standard deviation for patients offered problem-solving therapy compared to those receiving usual care (-0.36, 95% CI -0.61, -0.11). Three trials measured hopelessness, an item strongly correlated with suicidal ideation.^{69, 70, 93} Because these 3 trials used the same scale, the Beck Hopelessness Scale,⁹⁴ Townsend et al. calculate a Weighted Mean Difference. The groups receiving problem-solving therapy averaged approximately 3 points less on hopelessness scores at follow-up than those receiving standard care (-2.97 points; 95% CI, -4.81, -1.13). Two trials measured whether problems had improved (a dichotomous measure rated by assessors blinded to treatment).^{67, 68} Improvement in problems was more likely in those receiving problem-solving therapy compared to usual care (OR, 2.31; 95% CI, 1.29–4.13) (Table 8).

Older Adolescents and Adults: Additional Randomized Controlled Trials Involving Intermediate Outcomes

As shown in Table 9, Guthrie et al. measured suicidal ideation in their RCT by comparing 4 sessions of interpersonal psychotherapy delivered in the patient's home by nurse therapists to usual care.⁶¹ Patients were 18 to 65 years old. In an analysis that was not intention-to-treat, suicidal ideation (as measured by the Scale for Suicidal Ideation⁵¹) at 6-month follow-up showed a significantly lower degree of suicidal ideation (mean score 7.9 v. 12.8, 95% CI: -8.2 to -1.6, P = 0.0005). The authors had a priori identified a difference of 5 points as being clinically significant.

Koons et al measured suicidal ideation and depressive severity in their 6-month RCT of women veterans with borderline personality disorder.⁵⁷ In their analysis of those subjects completing treatment, DBT was superior to usual care in decreasing suicidal ideation as measured by the Scale for Suicidal Ideation⁵¹ (10 point mean decrease vs. 4 point mean decrease, P < 0.05 by two-way repeated measures analysis of variance [ANOVA]). The results for DBT's effect on depressive severity were inconsistent. As measured by the self-report Beck Depression Inventory,⁹⁵ DBT produced a significantly greater decrease in depressive symptoms than usual care (two-way ANOVA, P < 0.05). However, as measured by the interviewer-administered Hamilton Depressive Rating Scale,⁹⁶ there was no significantly greater decrease for DBT group vs. usual care by two-way ANOVA.

Montgomery et al. performed a 4-week cohort study comparing the antidepressants mianserin, amitriptyline, and maprotiline.⁹¹ This study was part of the work in defining the depressive measure used (the Montgomery-Asberg Depression Rating Scale, or MADRS) as being more sensitive to change for antidepressant clinical trials than the HAM-D. The authors found that suicidal ideation, as measured by the MADRS, was decreased by a significantly greater degree at study's end by mianserin compared to maprotiline (P < 0.01), and that there was a trend favoring mianserin over amitriptyline (P < 0.10). There was no difference among the 3 study drugs for the analogous “suicidal thoughts” on the HAM-D, and the overall quality of the study was poor.

Raj et al.'s previously described cohort study comparing 10 sessions of cognitive-behavioral counseling to routine medical treatment for patients admitted to an ICU following their first or second suicide attempt also measured the effect of the intervention on suicide ideation.⁸⁸ Assessing the difference in Scale for Suicidal Ideation (SSI) scores between baseline and 2 to 3 months post-discharge for the 2 groups, they found that those who received counseling had a substantially greater reduction in suicidal ideation than the usual care group (15.0 ± 7.79 vs. 2.75 ± 6.09, P ≤ 0.05).

Children 18 Years and Younger Exclusively: Randomized Controlled Trials Involving Intermediate Outcomes

In the Harrington et al. RCT reviewed above, the authors also compared the benefits of a home-based family intervention plus usual care to usual care alone for the degree of suicidal thoughts in adolescents 16 years of age and younger with DSH.⁴³ The authors found no effect on suicidal ideation measures for those completing treatment at 2- and 6-month follow-up. Subgroup analyses of adolescents without major depression at enrollment suggested that the intervention effectively decreased suicidal ideation at 2 and 6 months compared to the usual care group. Mean scores on the Suicidal Ideation Questionnaire at 2 months were 8.6 versus 32.8 (P < 0.01) and at 6 months 4.9 v. 21.6 (P < 0.01), respectively.

Wood et al. assessed the benefits of group therapy versus usual care for suicidal ideation in an RCT involving adolescents 12 to 16 years of age who were referred for mental health services as a result of DSH.⁴⁴ In an intention-to-treat analysis seven months post-randomization, the groups did not differ significantly in severity of suicidal ideation as measured by self-report on the 30-item Suicidal Ideation Questionnaire⁵² (mean difference from baseline was 47.3 v. 39.7; mean difference between interventions was 7.5 [95% CI, -18.8 to 33.9]).

Brent et al. compared cognitive behavior therapy (CBT), systemic behavior family therapy (SBFT), and individual nondirective supportive therapy (NST) for the treatment of adolescents between 13 and 18 years of age with major depressive disorders.⁹² Eligibility criteria did not require suicidal ideation or a history of self-harm. Primary outcome measures included suicidality and remission of major depressive disorder; analysis was intention-to-treat. The authors found that after 12 to 16 weeks of care, only 17% of adolescents randomized to the CBT group met the criteria for major depressive disorder, as did 32% receiving SBFT and 42% in the NST group (comparison among 3 groups revealed x² = 5.22; df = 2; P = 0.07; comparison between CBT and NST x² = 5.23, df = 1; P = 0.02). Suicidality decreased significantly for all 3 groups over the course of the study, with similar reductions seen across the 3 groups. These results suggest that more effective treatments for depression did not clearly lead to greater decreases in suicidality.

In the previously described cohort study by Rotheram-Borus et al. comparing an emergency room intervention of staff training, patient and family education, and 1 family therapy session to usual care in Latino females ages 12 to 18 years, the investigators also measured suicidal ideation and depressive symptoms.⁹⁰ At 18-month follow-up, the authors found no differences between groups in suicidal ideation. However, the authors did find benefit for the intervention group for depressive symptoms; the proportion of patients with Beck Depression Inventory scores⁹⁵ in the clinical range at 18 months was significantly lower for the intervention group (4.9% vs. 10.1%, P < 0.01).

Summary

Results from studies involving primarily older adolescents and adults are promising. In comparison to standard care, problem-solving therapy (meta-analyses showing benefit with improved mood, less hopelessness, and improvement in problems), interpersonal therapy (1 study demonstrating decreased suicidal ideation), DBT (with 1 trial demonstrating decreased suicidal ideation), and cognitive behavioral counseling (1 cohort study showing greater decrease in suicidal ideation) can improve intermediate outcomes in patients at high risk for suicidal ideation. Similar decreases in suicidal ideation for adolescent groups have not been shown. However, for adolescents 18 years of age and younger who have attempted suicide, 1 cohort study suggests that a brief emergency crisis intervention involving mother and daughter may decrease the number of patients with clinical relevant depressive symptoms at 18-month follow-up. No studies have been performed in patients recruited from primary care settings.