Remission, Recovery, Relapse and Recurrence Rates for Suicide Attempts and Non-Suicidal Self-Injury for Suicidal Youth Treated With Dialectical Behavior Therapy or Supportive Therapy

Abstract

Objective:

To evaluate rates of remission, recovery, relapse, and recurrence in suicidal youth who participated in a clinical trial comparing Dialectical Behavior Therapy (DBT) and Individual and Group Supportive Therapy (IGST).

Method:

Participants were 173 youth, ages 12–18, with repetitive self-harm (including at least one prior suicide attempt [SA]) and elevated suicidal ideation (SI). Participants received 6 months of DBT or IGST and were followed for 6-months post-treatment. The sample was 95% female, 56.4% White and 27.49% Latina. Remission was defined as absence of SA or non-suicidal self-injury (NSSI) across one 3-month interval; recovery was defined across 2 or more consecutive intervals. Relapse and recurrence were defined as SA or NSSI following remission or recovery. Cross-tabulation with χ² was used for between group contrasts.

Results:

Over 70% of the sample reported remission of SA at each treatment and follow-up interval. There were significantly higher rates of remission and recovery and lower rates of relapse and recurrence for SA in DBT versus IGST. Across treatments and time points, SA had higher remission and recovery rates and lower relapse and recurrence rates than NSSI. There were no significant differences in NSSI remission between conditions; however, DBT had significantly higher NSSI recovery rates than IGST for the 3–9, 3–12, and 6–12-month intervals.

Conclusion:

Results showed higher percentages of SA remission and recovery for DBT as compared to IGST. NSSI was less likely to remit than SA.

INTRODUCTION

Suicide is the second leading cause of death among 10 to 14-year-olds and the third leading cause of death among 15–24-year-olds in the United States.¹ Both suicide attempts (SA) and non-suicidal self-injury (NSSI) are risk factors for future SA and are prevalent among adolescents.^2,3 There are only a small number of treatments that have demonstrated efficacy in randomized controlled trials (RCTs) for decreasing SA, NSSI and/or self-harm (SH, defined as both SA and NSSI combined) in adolescents. Dialectical Behavior Therapy (DBT) is the only treatment shown to reduce SH in adolescents that has been replicated in two separate large-scale RCTs; and hence, that meets criteria for a “well-established” empirically supported treatment.^4–6 An additional small trial of DBT with adolescents found significantly fewer episodes of NSSI at the end of treatment in the DBT versus treatment as usual condition; however, a separate small feasibility trial found greater frequencies of SA and NSSI in DBT versus treatment as usual, but did not report if these findings were statistically significant or not.^7,8

Little is known about rates of remission, recovery, relapse, and recurrence of SA, NSSI, and SH in adolescents in response to suicide-focused treatments in general and to DBT specifically. This knowledge is essential to match adolescents to optimal treatment approaches, decide when to switch or augment treatments, and provide teens and parents with accurate estimates of prognosis. The primary purpose of the present work is to inform clinical practice by providing clinicians as well as adolescents and parents with estimates of remission, recovery, relapse, and recurrence rates in response to suicide-focused interventions that can be used to inform decisions and expectations about treatment options. To this end, in this paper we report remission, recovery, relapse, and recurrence rates for SA and NSSI in the Collaborative Adolescent Research on Emotions and Suicide (CARES) Study.⁵ The CARES Study is the largest RCT of DBT with adolescents to date and was a multisite trial with a sample of n = 173 adolescents with high suicide risk, defined as having at least one prior suicide attempt, ≥3 episodes of SH, and elevated suicidal ideation (SI). Participants were randomly assigned to receive 6-months of either DBT or individual and group supportive therapy (IGST). Both treatments used theoretically driven treatment manuals and were matched on the hours and modalities of treatment offered. In prior work, we reported significantly higher SH remission rates (defined as absence of SH) for DBT versus IGST, with 47% of participants in the DBT condition and 28% in the IGST condition reporting SH remission across the 6-month treatment period and 51% of youth in DBT and 32% in IGST reporting SH remission across the 6-month post-treatment follow-up period.⁵

The present study extends prior work by reporting relapse, recovery, and recurrence rates as well as remission rates, across all four longitudinal follow-up intervals included in the year-long study (0–3 months, 3–6 months, 6–9 months, and 9–12 months), and for SA and NSSI separately. Reporting remission, relapse, recovery, and recurrence rates every three months allows for a more detailed description of the length and timing of absence/presence of SA and NSSI across the treatment and follow-up periods. Moreover, examining SA and NSSI separately is important because they are distinct types of SH behaviors that serve different functions and can be differentiated based on the presence or absence of intent to die, as well as because SAs are a robust predictor of death by suicide, the most proximal behavior to death by suicide, and are associated with greater risk of death and/or serious injury than NSSI.^9–11

At present, there are no standardized definitions of remission, recovery, relapse, and recurrence of SA and NSSI within the suicide prevention literature. Hence, we based our definitions of these constructs on those used for major depression, given the close association between depression and suicidality.¹² For depression, consensus definitions of remission, recovery, relapse and recurrence were created by an expert task force in 1991.^13–17 These definitions have been widely used in studies on pediatric depression and similar definitions have been used for youth anxiety and eating disorders.^18–23 Remission is a period in which an individual is asymptomatic and no longer meets diagnostic criteria for the disorder and is distinct from response, which refers to a clinically meaningful decrease in symptoms or partial remission. Recovery is a sustained period of remission. Relapse is a return of symptoms during remission but before recovery and is distinct from recurrence, which is the appearance of an entirely new episode following a period of recovery.

Despite the longstanding use of these definitions, there is still considerable variability across studies regarding the criteria for remission of a major depressive episode (e.g., symptom free versus symptom reduction), the outcome measures used to assess remission, and the length of symptom-free time required for remission and recovery.^14,15,24 Similar difficulties in defining remission and recovery in adolescent anxiety and eating disorders have been noted.^19,20,25,26 Defining remission and recovery for SA, NSSI and/or SH presents additional challenges, because they are single behaviors versus multi-symptom diagnostic categories and there is no empirical consensus on the optimal amount SH-free time needed to reduce risk of future SH or death by suicide.

In the present study, we based our definitions of remission, recovery, relapse, and recurrence on those used for depression using the time frames of the CARES Study outcome assessments, which were conducted every three months during the 6 months of treatment (3 and 6-months post-baseline) and 6 months of follow-up (9 and 12-months post-baseline). We defined remission as the absence of SA or NSSI across one 3-month interval (i.e., 0–3, 3–6, 6–9, or 9–12 months). Recovery was defined as remission sustained across at least two consecutive 3-month intervals. Relapse was defined as SA or NSSI occurring during the 3-month interval immediately following a period of remission. We examined relapse of SA and NSSI at the end of the treatment (3–6-months) and post-treatment follow-up (9–12-months). Recurrence was defined as SA or NSSI following a period of recovery. We examined recurrence after recovery from 0–6 and 3–9-months. We predicted that youth who received DBT would have higher remission and recovery rates, as well as lower relapse and recurrence rates, for both SA and NSSI, than those who received IGST. We also examined non-remitters (e.g., participants who failed to obtain remission of SA or NSSI during active treatment (0–6 months), post-treatment follow-up (6–12 months) and across the entire year of the study (0–12 months). We expected that youth who received IGST would be more likely to be non-remitters at all time points than youth who received DBT.

METHOD

For a detailed description of the study design, see McCauley et al.⁵ and other prior publications from the CARES Study.^27,28 Here, we describe only the measures and procedures relevant to this report. The study was reviewed by each site’s local IRB. All youth and parents gave informed assent/consent to participate in the study.

Participants

Participants were recruited from January 2012 to August 2014 at four sites: University of Washington; Seattle Children’s Hospital; Harbor-UCLA Medical Center; and UCLA Medical Center. The final sample included N = 173 adolescents, ages 12–18, who were randomized to 6 months of either DBT or IGST. Participants completed a baseline assessment, as well as follow-up assessments every three months for a total of one year (see Figure S1, available online). Inclusion criteria were: ≥ 3 lifetime SH episodes, one within 12-weeks before screening; ≥ 1 lifetime SA; elevated past-month SI; and ≥ 3 borderline personality disorder (BPD) symptoms. Exclusion criteria were: IQ less than 70; court-ordered to treatment; primary psychosis, mania, anorexia, or other life-threatening condition; and youth not fluent in English or parent not fluent in English or Spanish. Mean age was 14.89 years (SD=1.47). The sample was 95% female, 56.4% White and 27.49% Latina (all Latina participants were female). Participants met criteria for the following DSM IV TR diagnoses: 83.81% depressive disorders, 54.1% anxiety disorder, 53% BPD. For a detailed description of sample demographics, as published previously,^5,28 see Table S1, available online.

Procedures

Both DBT and IGST were manualized, used similar training and adherence protocols, and offered equivalent treatment components (e.g., both individual and group therapy) to maximize internal validity. For a detailed description of the treatment approaches see Berk et al.²⁹ and McCauley et al.⁵ DBT included weekly individual psychotherapy, weekly multi-family group skills training, availability of 24/7 youth and parent telephone coaching, and a weekly therapist consultation team meeting. Family sessions were scheduled as needed, with a maximum of seven family sessions allowed. Suicide risk was monitored regularly; increased risk triggered use of the Linehan Suicide-Risk Assessment and Management Protocol (LRAMP).³⁰

IGST emphasized acceptance, validation, and feelings of connectedness/belonging and was non-directive. IGST included weekly individual psychotherapy, a weekly supportive therapy group for teens, and a weekly meeting for the therapists with the IGST supervisor. Therapists were available by phone during office hours; crisis numbers were provided for 24-hour coverage. Parent sessions were scheduled as needed, with a maximum of seven parent sessions allowed. Assessment and management of suicidal behavior followed the AACAP Practice Parameters.³¹

Therapist Training/Adherence

Therapists provided treatment in one study arm only and attended a multi-day training led by the treatment developer. Therapists participated in weekly cross-site training/meetings and weekly site team/consultation meetings (DBT) or group supervision (IGST). Treatment adherence was evaluated on randomly selected individual and group sessions using the DBT and IGST Adherence Scales respectively. Adherence was strong in both conditions: DBT (n=384 sessions), Mean 4.1 (≥4.0 considered adherent), SD 0.15; IGST (n=386 sessions), Mean 99.32 (>90% considered adherent), SD 3.64.⁵

Assessments

The first study assessment was conducted at baseline (pre-treatment), with follow-up outcome assessments conducted every three months during treatment (3 and 6-months post-baseline) and follow-up (9 and 12-months post-baseline). The Suicide Attempt Self-Injury Interview (SASII) was used to measure the number of SAs and NSSI episodes.³² The SASII has demonstrated strong reliability and validity; inter-rater reliability was maintained throughout the study at ICC ≥.80 at the item level.⁵

Randomization

Adaptive randomization balanced participants across conditions within sites based on age, number of prior SA, number of prior SH and psychotropic medication use.

Statistical Plan

Each variable (remission, recovery, relapse, and recurrence) was defined categorically (yes/no) over the respective time-window, with any non-zero occurrence of SA or NSSI leading to a categorization of “no” for remission and recovery and “yes” for relapse and recurrence. To account for attrition/missing data, we used Markov Chain Monte Carlo (MCMC) imputation methods to complete all missing data.^33–35 We ran 100 sets of MCMC imputations and conducted analyses using the average of the imputed data.³³ Treatment group contrasts were conducted using cross-tabulation methods with χ² for statistical significance. Attrition rates for each follow-up time point are reported in prior publications, with only a small number of participants missing all 4 follow-up evaluations and no significant differences between DBT (2/86 [2.3%]) and IGST (7/87 [8.0%]); Fisher exact test, p = .17 (see Figure S1, available online).⁵ For contrasts using variables with low n per cell, we replaced χ² test with Fisher’s exact test and Firth logistic regression which can accommodate both low prevalence as well as zero-cells.³⁶ All analyses were initially conducted as unadjusted using Rothman’s (1990) guidance that adjustments are not needed for multiple comparisons when study aims are exploratory.³⁷ We also conducted all analyses using the false discovery rate approach as sensitivity analyses adjusting for the number of tests and report these values (pFDR) as well as the unadjusted values in the tables.³⁸ We chose to use the FDR approach as opposed to a Bonferroni correction due to the intercorrelation between the multiple outcomes examined in this work.

RESULTS

Remission Rates

Remission rates for NSSI and SA are shown in Table 2. Remission rates for NSSI for the total sample ranged from a low of 32.4% during the 0–3-month interval to a high of 53.8% during the 6–9-month interval. The highest NSSI remission rate for DBT was 60.5% during the 6–9-month interval and the highest rate for IGST was 49.4% during the 9–12-month interval. There were no significant differences between treatment conditions at any time point, although percentage of youth in remission from NSSI was higher in DBT than IGST at all time intervals except the 0–3-month interval. Remission rates for SA were higher than those for NSSI, with more than 70% for the total sample reporting remission of SA at each time period. DBT yielded significantly higher rates of remission of SA than IGST during the 3–6-month and 9–12-month intervals, with 82.6% of youth in the DBT condition reporting remission at 3–6 months and 86.0% at 9–12 months, as compared to 62.1% and 65.5% for IGST.

Table 2.

Remission and Recovery Rates

Time Period	IGST	DBT	Total Sample	χ 2	p	p FDR
	% (freq)	% (freq)	% (freq)
NSSI Remission
0–3mo	34.5 (30/87)	30.2 (26/86)	32.4 (56/173)	0.36	0.55	0.58
3–6mo	37.9 (33/87)	50.0 (43/86)	43.9 (76/173)	2.54	0.11	0.18
6–9mo	47.1 (41/87)	60.5 (52/86)	53.8 (93/173)	3.08	0.08	0.15
9–12mo	49.4 (43/87)	55.8 (48/86)	52.6 (91/173)	0.71	0.40	0.47
NSSI Recovery
0–6mo	19.5 (17/87)	18.6 (16/86)	19.1 (33/173)	0.02	0.88	0.88
0–9mo	13.8 (12/87)	18.6 (16/86)	16.2 (28/173)	0.74	0.39	0.47
0–12mo	11.5 (10/87)	16.3 (14/86)	13.9 (24/173)	0.83	0.36	0.47
3–9mo	19.5 (17/87)	43.0 (37/86)	31.2 (54/173)	11.11	0.0009	0.0036
3–12mo	14.9 (13/87)	32.6 (28/86)	23.7 (41/173)	7.42	0.006	0.0171
6–12mo	29.9 (26/87)	46.5 (40/86)	38.2 (66/173)	5.07	0.02	0.0444
Suicide Attempt Remission
0–3mo	75.9 (66/87)	70.9 (61/86)	73.4 (127/173)	0.54	0.46	0.51
3–6mo	62.1 (54/87)	82.6 (71/86)	72.3 (125/173)	8.71	0.00	0.00
6–9mo	69.0 (60/87)	79.1 (68/86)	74.0 (128/173)	2.27	0.13	0.20
9–12mo	65.5 (57/87)	86.0 (74/86)	75.7 (131/173)	9.38	0.00	0.00
Suicide Attempt Recovery
0–6mo	54.0 (47/87)	64.0 (55/86)	59.0 (102/173)	1.76	0.18	0.26
0–9mo	44.8 (39/87)	59.3 (51/86)	52.0 (90/173)	3.63	0.057	0.11
0–12mo	40.2 (35/87)	58.1 (50/86)	49.1 (85/173)	5.55	0.019	0.0440
3–9mo	48.3 (42/87)	73.3 (63/86)	60.7 (105/173)	11.31	0.0008	0.0036
3–12mo	42.5 (37/87)	70.9 (61/86)	56.7 (98/173)	14.21	0.0002	0.0013
6–12mo	54.0 (47/87)	74.4 (64/86)	64.2 (111/173)	7.82	0.005	0.0167

Note: Remission and recovery are defined as absence of NSSI or SA during the given time interval. All results presented are for imputed data. For NSSI, with the observed data, for the 3–6-month interval, the remission rate for IGST was 40% as compared to 37.9% for the imputed data and for DBT, it was 56.9% for the observed data and 50.0% for the imputed data. At 9–12 months, for NSSI, the observed data yielded a 51.7% remission rate for IGST and 49.4% for the imputed data and, for DBT, a 62.0% remission for observed data and 55.8% for the imputed data. For SA, for the 3–6-month interval, the remission rate was 78.5% for observed and 62.1% for the imputed data for IGST and 90.3% for the observed data and 82.6% for the imputed data for DBT. At 9–12-months, the remission rate was 89.7% for the observed data and 65.5% for the imputed for IGST and 93.0% for the observed data and 86.0% for the imputed data for DBT. IGST= Individual and Group Supportive Therapy; DBT = Dialectical Behavior Therapy; fdr = false discovery rate, freq = frequency; NSSI = non-suicidal self-injury; SA = suicide attempt; mo = month, Significant p-values are bolded.

Recovery Rates

As shown in Table 2, recovery rates for both NSSI and SA were lower for when the periods of recovery were longer. For the total sample, NSSI recovery rates ranged from a high of 38.2% during the 6–12-month interval to a low of 13.9% for the 0–12-month time interval. There were no significant between-group differences in recovery rates for NSSI between DBT and IGST for the 0–6-, 0–9-, and 0–12-month intervals; however, there were significant between group differences in recovery rates during the 3–9, 3–12, and 6–12-month intervals, with rates for DBT ranging from 43.0% to 46.5% and IGST from 14.9% to 29.9%. For SA, for the total sample, recovery rates ranged from a high of 64.2% during the 6–12-month interval to a low of 49.1% during the 0–12-month interval. There were significant differences in SA recovery rates between DBT and IGST at the 0–12, 3–9, 3–12, and 6–12 month intervals, with rates in the DBT condition ranging from 74.4% to 58.1% and IGST rates ranging from 54.0% to 40.2%. Across both treatments and all time periods, recovery rates were higher for SA than NSSI.

Relapse and Recurrence Rates

As shown in Table 3, for the total sample, relapse and recurrence rates for NSSI ranged from a high of 41.1% to a low of 24.1%. There was a marginally significant difference between conditions in recurrence during the 6–12-month follow-up period following recovery during the 0–6-month treatment period, with 12.5% of youth in DBT reporting recurrence of NSSI versus 41.2% of youth in IGST; however, this difference was no longer marginally significant using the adjusted p-value (p^FDR = .13). For SA, relapse and recurrence rates ranged from a high of 19.7% to a low of 6.8%. There were significantly higher rates of SA relapse for IGST (28.8%) than for DBT (9.8%) during the second half of treatment (3–6 months) and during the second half of the follow-up phase (9–12 months; IGST: 21.7%; DBT: 5.9%). There was also a significant difference in SA recurrence rates using the unadjusted p value, with 9.1% of youth who received DBT reporting recurrence during the follow-up phase after recovery during treatment as compared to 25.5% of youth in IGST; however, this result was no longer significant using the adjusted p value (p^FDR = .07).

Table 3.

Relapse and Recurrence Rates

Time Period	IGST % (freq)	DBT % (freq)	Total Sample % (freq)	χ 2	p	pFDR
NSSI Relapse
Remission at 0–3mo/relapse at 3–6mo	43.3 (13/30)	38.5 (10/26)	41.1 (23/56)	0.14	0.81	0.93
Remission at 6–9mo/relapse at 9–12mo	36.6 (15/41)	21.1 (12/52)	29.0 (27/93)	2.03	0.15	0.20
NSSI Recurrence
Recovery 0–6mo/any recurrence	41.2 (7/17)	12.5 (2/16)	27.3 (9/33)	3.47	0.065	0.13
Recovery 3–9mo/recurrence 9–12mo	23.5 (4/17)	24.3 (9/37)	24.1 (13/54)	0.004	0.95	0.95
Suicide Attempt Relapse
Remission at 0–3mo/relapse at 3–6mo	28.8 (19/66)	9.8 (6/61)	19.7 (25/127)	7.20	0.007	0.036
Remission at 6–9mo/relapse at 9–12mo	21.7 (13/60)	5.9 (4/68)	13.3 (17/128)	6.90	0.009	0.036
Suicide Attempt Recurrence
Recovery 0–6mo/any recurrence	25.5 (12/47)	9.1 (5/55)	16.7 (17/102)	4.94	p=0.026	0.069
Recovery 3–9mo/recurrence 9–12mo	11.9 (5/42)	3.2 (2/63)	6.8 (7/105)	3.09	p=0.079	0.13

Note: Relapse is defined as presence of NSSI or SA following a period of remission. Recurrence is defined as presence of NSSI or SA following a period of recovery. IGST= Individual and Group Supportive Therapy; DBT = Dialectical Behavior Therapy; fdr = false discovery rate, freq = frequency; NSSI = non-suicidal self-injury; SA = suicide attempt; mo = month, Significant p-values are bolded.

Nonremitters

As shown in Table 4, the rate of non-remission of NSSI across the entire 12 months of the study was 19.1% for the total sample, with 42.8% reporting no remission during active treatment (0–6 months) and 31.8% reporting no remission during follow-up (6–12 months). There were no significant differences between conditions in the percentage of youth who did not report any remission of NSSI during treatment, follow-up, and/or across both treatment and follow-up. For SA, in the total sample, only 8.7% of youth reported no period of remission of SA across the entire study (0–12 months), 13.3% were non-remitters during active treatment, and 14.5% did not remit during follow-up. There were no differences between treatment conditions for the percentage of youth who did not remit during treatment; however, there were marginally significant findings showing greater rates of non-remission for IGST versus DBT during the follow-up phase and across the whole study using unadjusted p-values; these differences were no longer significant using the adjusted values.

Table 4.

Nonremitters by Time Period and Treatment Condition

Time Period	IGST % (freq)	DBT % (freq)	Total Sample % (freq)	χ 2	p	pFDR
No NSSI Remission
Active Treatment (0–6M)	47.1 (41/87)	38.4 (33/86)	42.8 (74/173)	1.35	0.24	0.42
Follow-up (6–12M)	33.3 (29/87)	30.2 (26/86)	31.8 (55/173)	0.19	0.66	0.79
All time points (0–12M)	18.4 (16/87)	19.8 (17/86)	19.1 (33/173)	0.05	0.88	0.98
No Suicide Attempt Remission
Active Treatment (0–6M)	16.1 (14/87)	10.5 (9/86)	13.3 (23/173)	1.19	0.28	0.56
Follow-up (6–12M)	19.5 (17/87)	9.3 (8/86)	14.5 (25/173)	3.67	0.056	0.19
All time points (0–12M)	12.6 (11/87)	4.7 (4/86)	8.7 (15/173)	3.49	0.062	0.19

Note: Remission is defined as absence of NSSI or SA during the given time interval. IGST = Individual and Group Supportive Therapy; DBT = Dialectical Behavior Therapy; fdr = false discovery rate, freq = frequency; NSSI = non-suicidal self-injury; SA = suicide attempt; mo = month, Significant p-values are bolded.

DISCUSSION

We examined remission, recovery, relapse, and recurrence of SA and NSSI in a sample of youth at high risk for suicide, in 3-month intervals, across a total of 12 months, including 6 months of treatment with either DBT or IGST and 6 months of post-treatment follow-up. This is the first study to our knowledge to evaluate these indices of treatment outcome for SA and NSSI in youth. For the total sample, rates of SA remission were high, with over 70% of the sample reporting remission at each of the treatment and follow-up intervals. SA recovery rates were lower than remission rates and decreased with the amount of time intervals included, with a high of 64% of the total sample in remission from SA in the 6–12-month interval and a low of 49.1% across the entire 0–12-month period. Remission and recovery rates were observed to be lower for NSSI than for SA across all time intervals and treatment conditions. NSSI remission rates for the total sample ranged from 32.4 – 53.8% and recovery rates ranged from 13.9 (0–12 months) to 38.2% (6–12 months), with longer periods having lower rates. Rates of remission and recovery showed different patterns over time for SA and NSSI. For SA, remission rates across conditions were high during the first three months of treatment and continued to improve over subsequent time periods for DBT whereas IGST rates worsened over time. In contrast, for NSSI, remission rates were lowest during the first three months of treatment and increased over subsequent time intervals in both conditions, with higher percentages of remission and recovery in DBT than in IGST. These findings suggest that NSSI may be slower and less likely to remit than SA, perhaps due to the higher frequency and more habitual nature of NSSI behaviors.

For SA, with both treatment conditions showing high rates of remission of SA during the 0–3-month interval, our findings suggest that the first three months of a suicide-focused treatment may have an initial protective effect. Potential reasons for this may include: the intensive safety planning that typically occurs at the beginning of treatment with suicidal individuals; increased hopefulness in relation to starting a new treatment; and/or the lower frequency of SAs as compared to NSSIs. Our results inform questions about optimal treatment length by suggesting that initial SA remission at the 3-month time point may not be sustained and thus may be too soon to stop treatment, particularly given that a subsequent suicide attempt may lead to serious injury or death. These results parallel our prior work on longitudinal trajectories of treatment response in the CARES study sample, which showed that all youth showed some improvement in total SH over the first three months of treatment and then sharply diverged into responder and non-responder groups over the remaining months of treatment and follow-up.³⁹ Further research is needed to verify and understand these findings, as well as to identify predictors of relapse of SA during treatment after an initial period of remission.

Results showed significantly higher rates of remission and recovery and lower rates of relapse and recurrence for SA in DBT as compared to IGST. In particular, youth who received DBT reported significantly higher remission rates for SA during the 3-to-6-month interval (middle to end of treatment) and the 9-to-12-month interval (middle to end of the follow-up period), as well as significantly higher rates of recovery from 3–9, 3–12, and 6–12-month intervals. Youth in DBT versus IGST also reported significantly higher rates of recovery for SA across all four study assessment intervals (0–12 months). Participants in the IGST condition also had significantly higher rates of relapse (re-emergence of SA following a remission) of SA during the 3-to-6-month interval and during the 6–9-months interval. Finally, youth in IGST had higher rates of recurrence of SA following recovery during the treatment period than those in the DBT condition. DBT also had significantly higher rates of NSSI recovery for the 3–9-, 3–12-, and 6–12-month intervals, but there were no significant differences between conditions for NSSI remission at any time point.

Notably, remission and recovery rates for SA for DBT were high, with 83% reporting remission at the end of treatment and 86% reporting remission at the end of post-treatment follow-up, 64% reporting recovery across 0–6 months, and 74% recovery across the 6–12-months. These high remission and recovery rates suggest that DBT offers substantial protection against repeat and potentially fatal SAs and that most suicidal youth who are treated with DBT may attain at least one period of remission of SA during treatment and follow-up. High post-treatment remission rates also suggest that the benefits of DBT may persist after treatment has been completed. Our prior work showing that improvements in youth emotion regulation during treatment mediated the association between DBT and SH remission during follow-up in the CARES Study sample suggests that emotion regulation skills may be a critical treatment component for sustained SA remission.⁴⁰ Other reasons DBT may have led to greater reductions in SA than IGST include the focus on commitment to eliminating SA, close tracking of SA and NSSI on the weekly diary card and the skills-focused, directive nature of DBT; however, additional work is needed to isolate the active ingredients in DBT. Taken together, these results highlight the effectiveness of DBT in reducing SAs and underscore the importance of increasing the availability and accessibility of DBT for suicide prevention in youth.

Results for IGST, with 76% of youth reporting remission of SA at 3-month follow-up and approximately two-thirds of youth in the IGST condition reporting remission at 3–6, 6–9, and 9–12-month intervals, suggest that supportive therapy also offers benefit for reducing suicide risk. For both DBT and IGST, rates of sustained recovery of SA (i.e., remission over multiple consecutive time intervals) decreased as more time intervals were included, with 58% of youth reporting remission across all four assessment intervals in DBT and 40% in IGST, suggesting there is still room for improvement in both treatments and that a waxing and waning course of self-harm behaviors may be more common than sustained recovery. Hence, it may make sense clinically to continue risk assessment and safety planning during periods of remission and recovery until it is clear if a sustained recovery will be achieved.

There were no differences between DBT and IGST in relapse or recurrence rates for NSSI. However, for SA, there were significantly higher relapse rates for IGST than DBT during the second half of treatment (3–6 months) and follow-up (9–12 months) and for recurrence following remission during treatment. The percentage of non-remitters was low, with only 8.7% of youth showing no remission of SA at any time point and 19.1% showing no remission of NSSI at any time point. There were no significant differences between conditions for non-remitters at any time point, likely due to the small sample size for these calculations. The low percentage of non-remitters is encouraging and suggests that most self-harming youth will experience a period of remission of SA and/or NSSI at some point during treatment with DBT or supportive therapy.

We created definitions of remission, recovery, relapse, and recurrence of SA and NSSI based on the length of treatment and follow-up periods used in the CARES trial. However, consensus definitions of these constructs have not yet been developed and studies that use different definitions and/or time frames may yield different results. We did not examine treatment response (defined as a clinically meaningful reduction in symptoms or partial remission) given the difficulties of applying this construct to SA and NSSI. Because SAs are single behaviors that occur with low frequency, even in this very high-risk sample, there was a restricted range to measure percent reduction in SAs. It is also unclear if reduction versus remission of SA is clinically meaningful given the potential for death/serious injury with any SA. Similarly, for NSSI, there is no data to guide what percentage of reduction is clinically significant. Given the lower rates of sustained remission of NSSI, as compared to SA, further research is needed to determine what constitutes a clinically meaningful and safe reduction in NSSI if remission cannot be obtained.

Limitations to generalizability include the predominately female and high-risk sample and results may not generalize to males or gender-expansive youth and/or to youth without histories of SA and repetitive SH. Additionally, our analyses were based on remission status at each time point within the entire sample, regardless of a given participant’s remission status at other time points; and hence, do not reflect trajectories of remission/recovery over time (e.g., how many participants who were in remission remained in remission at the next time point versus relapsed).

All analyses reported in this work were exploratory and therefore were not pre-specified. We included both adjusted and unadjusted p values for all comparisons. Adjusted analyses using the false discovery rate technique found that all significant results remained significant at the 0.05 level of significance (except for SA recurrence during the follow-up period after recovery during treatment). Replication of these results in other studies and samples is needed to verify these findings.

In this paper, we used multiple imputation to complete all missing data. Compared to the previously published outcomes reported in McCauley et al.,⁵ 2018, for which we used the observed data only, the imputed data yielded slightly more conservative rates of remission and recovery. The imputation techniques of many software algorithms are based on a continuous distribution for the missing outcome. Sensitivity of imputation techniques for count data such as our SA and NSSI measures, is an area of active research as seen in Liu and Reiter’s (2022) recent methods paper on this topic.⁴¹ Future research may benefit from these advanced techniques of imputing count data once they’re more readily accessible to researchers.

In summary, these results highlight the unique contribution of examining SA and NSSI separately by showing differences in rates and patterns of remission and recovery over time, extending our prior findings on SH remission rates and trajectories of treatment response.^5,28,39 At present, DBT is the only well-established treatment for decreasing SH in youth.⁶ Results of this study, showing very high remission and recovery rates for SA for youth in the DBT condition, further underscore the value of DBT as a first line treatment for youth at high risk for suicide. To the best of our knowledge, this is the only study of treatment approaches for suicidal youth that has defined and reported rates of remission, recovery, relapse, and recurrence for SA and NSSI. The lack of empirically based consensus definitions of these constructs is a critical gap within the suicide prevention literature and additional work on this topic is urgently needed.

Table 1.

Pretreatment Demographic, Self-injury and Diagnostic Data by Condition^a,b

Variable	DBT Group	IGST Group	Total	Statistic and p value
	(N = 86)	(N = 87)	(N=173)
Female	95.30% (N = 82)	94.19% (N = 81)	94.8% (N=163)	χ2 = 0.13, p=0.72
Age	14.77 (SD=1.50)	15.04 (SD=1.43)	14.89 (SD=1.47)	t (169) = 1.24, p=0.22
Race/Ethnicity
White	58.14%(N=50)	55.29% (47)	56.39 (N=97)
Native American	1.16% (N=1)	0% (N=0)	.58 (N=1)
African American	8.14% (N=7)	5.88% (N=5)	7.02 (N=12)	χ2 (5) = 2.84, p=0.72
Asian American	4.65% (N=4)	7.06% (N=6)	5.85 (N=10)
Other	1.16% (N=1)	3.53% (N=3)	2.34 (N=4)
Hispanic	26.7% (N=23)	28.24% (N=24)	27.49 (N=48)
Parental Marital Status
Married	57.14% (N=44)	52.05% (N=38)	54.67% (N=82)
Single, divorced, or separated	40.26% (N=31)	43.84% (N=32)	42.00% (N=63)	χ2 (3) = 0.68, p=0.88
Widowed	1.30% (N=1)	2.74% (N=2)	2.00% (N=3)
Other	1.30% (N=1)	1.37% (N=1)	1.33% (N=1)
Parental Education
Less than high school	8.86% (7)	6.76% (5)	7.84% (12)
High school graduate or GED	12.66% (10)	12.16% (9)	12.42% (19)	χ2 (3) = 0.54, p=0.91
Some college or tech. school	16.46 (12)	20.274% (15)	18.30% (28)
College graduate	62.03% (49)	60.81% (45)	61.44% (94)
Income
Below $15,000	11.94% (8)	10.14% (7)	11.03% (15)
$15,000-$29,999	5.97% (4)	7.25% (5)	6.62% (9)	χ2 (7) = 6.95, p=0.43
$30,000-49,999	11.00.(8)	23.9% (17)	17.4% (25)
$50,000+	71.2% (52)	59.2% (42)	65.3% (95)
SIQ-J Score	57.88 (SD = 17.01)	56.23 (SD = 15.37)	57.06 (SD =16.18)	t (171) = 0.67, p=0.51
Lifetime Suicide Attempt –SASII
1 attempt	39.5%(N=34)	42.5% (N=37)	41.0% (N=71	χ2 (1) = 0.16, p=0.69
> 1 attempt	60.5% (N=52)	57.5% (N=50)	59.0% (N=102)
Lifetime NSSI - SASII	26.29 (SD = 43.06)	29.14 (SD = 52.63)	26.32 (SD=47.19)	t (171) = 0.34, p=0.73
Depressive Disorders	79.1% (N=68)	88.5% (N=77)	83.81% (N=145)	χ2 (1) = 2.84, p=.10
Anxiety Disorders	48.8% (N=42)	59.3% (N=51)	54.1% (N=93)	χ2 (1) = 2.11, p=.17
Eating Disorders c	1.16% (N=1)	0	.68% (N=1)	χ2 (1) = 1.00, p=0.32
Borderline Personality Disorder	50% (N=43)	56.3% (N=49)	53.2% (N=92)	χ2 (1) = 0.69, p = .45
CBCL Externalizing T score	64.68 (SD = 11.27)	62.01 (SD = 16.11)	66.05 (SD=8.45)	t (166) = 1.25, p = .21
DUSI d
Average problem density score	22.73 (SD=24.77)	21.51 (SD=26.43)	22.12 (25.55)	t (165) = .31, p=.76

Note: Adapted from McCauley et al.,⁵ 2018 (https://doi.org/10.1001/jamapsychiatry.2018.1109). Reproduced with permission from JAMA Psychiatry. 2018. 75(8): 777–785. Copyright© (2018) American Medical Association. All rights reserved. CBCL = Child Behavior Checklist; DBT= dialectical behavior therapy; DUSI = Drug Use Screening Inventory; IGST = individual and group supportive therapy; SASII = Suicide Attempt Self-Injury Interview; SIQ-JR = Suicidal Ideation Questionnaire Junior.

^aNote: Data are presented as number (percentage) of participants unless otherwise indicated.

^bPsychiatric diagnosis was established for past year and current status; current status is reported in Table 1.

^cA total of 143 of 173 participants completed the eating disorders module of the Schedule for Affective Disorders and Schizophrenia for School-Aged Children because of a protocol change.

^dDUSI scores reflect an overall past month problem density score, ranging from 0% to 100%.