Acute Phase Treatment Outcomes of Electroconvulsive Therapy in Adolescents and Young Adults

Materials and Methods

Statistical Analysis

We calculated univariate statistics (e.g., means/standard deviation, percent) of the patient demographic and clinical characteristics, and ECT parameters, at baseline. We also compared these characteristics to the excluded population (i.e., adolescents and young adults who received ECT but who had incomplete information at baseline or the interim CMI assessment after treatment five) to examine if the excluded population differed from our study population in these observable characteristics.

Analyses were then conducted in the study population, to assess change from baseline to the mid-acute phase (i.e., after the fifth ECT treatment) follow-up assessment in the five BASIS-24 domains using linear mixed models that included the effect of time to estimate the mean change and incorporated repeated measures of the BASIS-24 domains on each subject. To determine whether changes in the BASIS-24 domains were associated with screening positive for an alcohol or drug use disorder, a binary indicator for screening positive for a SUD was added to the linear mixed models that included the effects of time; in addition, the inclusion of an interaction between time and the binary indicator provided a formal comparison of whether the pattern of changes in the means of the BASIS-24 domains differed by those who screened positive for SUD versus those who screened negative.

All statistical analyses were conducted using SAS, version 9.4.

Results

The total number of adolescents and young adult patients who received ECT during the study period was N=292. Among these, N=102 (34.9%) were excluded from the analysis due to missing information, that is, they were missing either the initial (baseline) CMI assessment or the repeat BASIS-24 after the fifth ECT treatment. Thus, our study population consisted of the remaining N=190. During the study period, approximately 2,890 inpatients at our institution aged 16–25 years were treated, with approximately 6% treated with ECT. The study population and those excluded from analysis did not differ in a statistically significant way with regard to age, sex, race, and level of care where receiving ECT (i.e., inpatient, residential or ambulatory care) (data not shown). However, they did differ in terms of diagnosis, with patients diagnosed with depressive disorders less likely to be in the excluded population, while those with bipolar disorder or primary psychotic disorder were more likely (p=0.002). Those who were excluded from the study were also more likely to start with bilateral ECT rather than unilateral ECT (p < 0.001). The differences in diagnosis and ECT placement and pulse parameters suggest that the included sample had less severe illness compared to those who were excluded. Because incomplete information about SUD screening and BASIS-24 on admission was part of the exclusion criteria, data on these two variables were available for only approximately two thirds of the excluded sample. When compared to the included sample, there were no significant differences in rates of positive SUD screens and in mean BASIS-24 scores in any of the subdomains.

In the study population, the mean age was 21.0 years, 16 patients were under age 18, and 51.6% were female (Table 1). Sixty-four percent (N=122) screened positive for a substance use disorder. There was a variety of psychiatric diagnoses among the population of patients who received ECT, the most prevalent being depressive disorders (62.1%) and psychotic disorders (16.5%). The majority (67.9%) of patients initiated their ECT course while inpatient. Baseline BASIS-24 scores ranged from a mean (standard deviation) of 0.61 (0.89) for psychosis and 2.69 (0.88) for depression/functioning.

ECT was associated with significant clinical improvement in all five BASIS-24 domains in mid-acute phase treatment. Mean domain decreases (i.e., improvements) were found in depression/functioning (−0.84 ± 0.070, p<0.001; t=−12.05, df=189), interpersonal relationships (−0.36 ± 0.064, p<0.001; t=−5.59, df=189), self-harm (−0.65 ± 0.071, p<0.001; t=−9.05, df=189), emotional lability (−0.35 ± 0.067, p<0.001; t=−5.25, df=189), and psychosis (−0.17 ± 0.049, p<0.001; t=−3.58, df=189) (Table 2).

Compared to screening negative for a SUD, screening positive for a SUD was associated with greater self-reported mid-acute phase clinical improvement in several BASIS-24 domains: depression/functioning (−0.99 vs. −0.62, p=0.009; t=−2.63, df=188), interpersonal relationships (−0.47 vs. −0.20, p=0.045; t=−2.02, df=188), and emotional lability (−0.46 vs. −0.19, p=0.044; t=−2.03, df=188) (Table 3).

Discussion

To our knowledge, this is the largest case series of ECT in adolescents and young adults published to date. Our data show that acute phase treatment in transitional age youth who have varied psychiatric disorders was associated with significant improvement, as demonstrated by patient self-reported outcomes in the BASIS-24. This is consistent with prior literature from smaller case series in adolescents and young adults with psychotic disorders, depression, and catatonia^17,26.

Additionally, we found that screening positive for co-occurring SUD was common in our sample (60%), and associated with greater improvement in acute phase ECT outcomes in BASIS-24 subdomains of self-reported depression/functioning, interpersonal relationships, and emotional lability, compared to screening negative.

Possibly, our finding of improved outcomes in adolescents and young adults who screened positive for SUD is an artifact of abstinence from substance use during ECT treatment, rather than a direct effect of ECT on these clinical domains. Inpatients would not have access to alcohol or drugs during ECT treatment and outpatients at our institution are asked to cease the use of all substances during their ECT treatment course. In particular, opioid and cocaine use are a contraindication to ECT. Patients who cannot abstain from medical or other marijuana are asked to minimize their use to avoid difficulty evaluating efficacy/side effects of ECT. Anecdotally, we have observed that marijuana use during ECT tends to aggravate memory issues, both anterograde and retrograde. Alcohol use can affect seizure threshold and misuse can affect the safety of the anesthesia. In addition, adolescents and young adults may have lower seizure thresholds given their age and thus are at additional risk of prolonged seizures with alcohol use⁹. Also possible is that ECT itself (independent of reduced substance use/abstinence) may improve outcomes among adolescents or young adults, as it has been associated with activation of the dopamine system and enhanced modulation of dopamine transmission²⁷. Alteration of the dopaminergic system is also implicated in substance use disorders, which may explain why those who screened positive for substance use disorders had significantly greater improvement from ECT. Further research will be needed to better clarify this relationship and how the dopaminergic system may be altered in those with substance use disorders. Though our finding differs from reports in the literature of adult patients where co-occurring SUD in general is associated with worse clinical outcomes for patients with psychiatric illness²⁸, an emerging ECT literature in adults has examined the association between SUD and ECT response and has found mixed results. Aksay and colleagues found that co-occurring substance use disorders was a predictor of worse clinical response to ECT²⁹, whereas Moss and Vaidya found that a history of co-occurring alcohol use disorders was a positive predictor of response to ECT³⁰. More research is needed to clarify the role substance use plays in the recovery of adolescents and young adults who receive ECT and determine whether confounding factors may have influenced our results.

There are some limitations worth noting in this study. First, our study was conducted at a single site and, rather than report on a psychiatric sample described by structured diagnostic interviews, we instead used a self-report symptom tool (the BASIS-24). However, the BASIS-24 is a well-validated tool to measure psychiatric symptom severity that has been validated in heterogeneous psychiatric diagnostic populations. Further, our baseline BASIS-24 domain means are similar in severity to means in a national sample of patients hospitalized for psychiatric disorders²², thus demonstrating the generalizability of our sample from a national inpatient sample. Therefore, while we are not able to report our study results based on structured diagnostic research interviews, our results do reflect symptom outcomes in a usual care sample which better reflects the heterogeneity of adolescents and young adults who receive ECT. Additionally, we examined outcomes at mid-acute phase (i.e., after five treatments, approximately two weeks), which is shorter than the typical clinical duration of acute phase ECT. However, the adult ECT research finds that younger age is associated with earlier response²³, most patients respond to ECT within five treatments²⁴, and early response is associated with sustained response and remission^24,25. In adolescents, a longer course of ECT may be required before a response is clinically apparent³¹, therefore, while we expect that our results are a conservative estimate of clinical outcomes, measuring ECT response within five treatments is a clinically meaningful outcome. Third, the substance use measure included was a screen, not a diagnostic tool, assessing for problematic alcohol or drug use in the past thirty days and did not quantify substances of use or quantity of use. In addition, thirty-five percent of the adolescents and young adults who received ECT during the study period were excluded from our sample because they either did not have complete baseline information/assessments in the CMI, or did not complete a repeat assessment after the fifth treatment. We found that the patients who were excluded were similar in several observable demographic and clinical characteristics, but differed based on two characteristics consistent with illness severity diagnosis (i.e., individuals with bipolar disorder and psychotic disorders) and ECT placement and pulse parameters. These differences may be largely explained by the fact that patients were excluded based on missing data from computerized patient self-reported assessments, which may be more difficult for patients who are severely ill to complete due to symptoms including anxiety, agitation or psychosis. Our exclusion criteria were incomplete information about SUD screening status and BASIS-24 on admission or discharge; therefore due to data missingness, we were unable to provide definitive comparisons of the excluded versus included population on these characteristics. Thus, there is the possibility of a selection bias between those who were excluded due to incomplete assessments compared to those included in the study sample. Future study is needed to understand if there are differential clinical effects of ECT based on patient diagnosis and/or illness severity. Another limitation to note is that our study did not include a non-intervention control group and therefore we cannot rule out that the patients might have improved (or improved similarly) absent the ECT intervention. That said, typically ECT is prescribed after several rounds of ineffective prior treatment and/or in treatment-resistant individuals, making a “placebo effect” in these ECT-treated adolescents and young adults less likely. Also, we note that our assessment of the association between clinical improvement and screening positive for co-occurring SUD was not subject to this limitation.

This study provides new and important information about a large sample of adolescents and young adults with severe mental illness who receive ECT in a large hospital-based usual-care sample. ECT was associated with clinical improvement. Furthermore, adolescents and young adults who screened positive for a SUD had greater improvement in clinical outcomes, compared to those who screened negative. More research is needed to clarify further which adolescents and young adult patients characteristics may be associated with differential ECT outcomes, and whether the differential improvement in certain domains for patients who screen positive for co-occurring SUD may be due to confounding factors (including, temporary abstinence from substances during treatment).