Association between pulse width and health-related quality of life after electroconvulsive therapy in patients with unipolar or bipolar depression: an observational register-based study

Abstract Aims To examine the association between pulse width and HRQoL measured within one week after electroconvulsive therapy (ECT) and at six-month follow-up in patients with unipolar or bipolar depression. Methods This was an observational register study using data from the Swedish National Quality Registry for ECT (2011–2019). Inclusion criteria were: age ≥18 years; index treatment for unipolar/bipolar depression; unilateral electrode placement; information on pulse width; EQ-5D measurements before and after ECT. Multiple linear regressions were performed to investigate the association between pulse width (<0.5 ms; 0.5 ms; >0.5 ms) and HRQoL (EQ-5D-3L index; EQ VAS) one week after ECT (primary outcome) and six months after ECT (secondary outcome). Results The sample included 5,046 patients with unipolar (82%) or bipolar (18%) depression. At first ECT session, 741 patients (14.7%) had pulse width <0.5 ms, 3,639 (72.1%) had 0.5 ms, and 666 (13.2%) had >0.5 ms. There were no statistically significant associations between pulse width and HRQoL one week after ECT. In the subsample of patients with an EQ-5D index recorded six months after ECT (n = 730), patients receiving 0.5 ms had significantly lower HRQoL (−0.089) compared to <0.5 ms, after adjusting for demographic and clinical characteristics (p = .011). The corresponding analysis for EQ VAS did not show any statistically significant associations. Conclusion No robust associations were observed between pulse width and HRQoL after ECT. On average, significant improvements in HRQoL were observed one week and six months after ECT for patients with unipolar or bipolar disease, independent of the pulse width received.


Introduction
Major depressive disorder is one of the leading causes of disability, with more than 250 million individuals affected worldwide [1].Having depression, which could be unipolar or bipolar, has a substantial negative impact on the individual in terms of morbidity, impaired function and health-related quality of life (HRQoL), and mortality [2].Electroconvulsive therapy (ECT) is a treatment in which electric current is applied to the scalp to provoke a seizure [3].ECT has shown to be an effective treatment for patients with severe depression [4,5], but is also associated with side effects such as temporary memory loss [4].
In ECT practice, different treatment modalities can be adjusted to maximize the treatment effect while reducing the cognitive side effects [6,7].One such adjustment concerns the pulse width [8].The transition from the original sine wave stimulation (≥8.33 ms) to brief pulse width (0.5-2 ms) has revealed benefits in terms of reduced cognitive side effects [9].It has been hypothesized that the use of ultra-brief pulse width (<0.5 ms) may induce a sufficient seizure to gain a treatment effect while also reducing the cognitive side effects [8,9].However, a key concern with adopting ultra-brief pulse width is that it may perform worse in terms of antidepressant efficacy.The relationship between different pulse widths and treatment effects is not fully understood, and there is little evidence from previous research on this particular area [10].
The utilization rates and treatment practices of ECT vary internationally [3], and the scientific literature presents inconsistent findings regarding the efficacy (e.g.speed of response and remission) of different pulse widths [7,11,12].The synthesis of findings from three literature reviews is further complicated by the use of different thresholds to define brief and ultra-brief pulse [7,11,12].Through a better understanding of the association between pulse width and health effects, treatment modality adjustments could be made to optimize the treatment effect for patients.
The patients' own assessment of their HRQoL may capture the impact of both treatment effects and adverse effects into one measure.A few studies have identified improvements in patients' HRQoL after ECT [13][14][15][16].Yet, to our knowledge, only one study (including 355 patients with major depressive disorder or bipolar disorder) has included information on pulse width in the analysis of potential factors that may influence patients' HRQoL after ECT [13].However, since this Australian study also examined the possible influence of different electrode placement, the subsamples for each of the placements were relatively small (unilateral placement, n = 76).By using national register data for a large sample of patients who have received ECT for unipolar or bipolar depression, the current study adds to the existing knowledge of real-world outcomes, as assessed from the patient perspective.
The aim of this study was to examine the association between pulse width and health-related quality of life measured within one week after ECT and at six-month follow-up in patients with unipolar or bipolar major depression.

Study design
The study was an observational register study.A study protocol was developed prior to data collection [17], and the reporting follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement [18].

Data source
Data was obtained for patients registered in the Swedish National Quality Register for ECT (Q-ECT), with linkages to the longitudinal integrated database for health insurance and labour market studies (LISA) (Statistics Sweden), and the National Patient Registry and the Swedish Prescribed Drug Registry (the National Board of Health and Welfare).
The Q-ECT was established in 2011 to enable monitoring of the Swedish clinical guidelines for ECT and is used for quality assurance and research [19].Q-ECT collects data including individual-level information on patient characteristics, symptoms, indications, ECT modalities, and side effects.A recently published study examining the data quality in Q-ECT demonstrated high accordance compared to patient records (i.e.89% or higher) for registration of diagnosis, treatment dates, and disease-specific rating scales [20].While Q-ECT is a non-mandatory register, there is high coverage of hospitals across Sweden and of patients contributing with data to the register.Of all patients treated with ECT in Sweden, the registry coverage increased from 79% in 2012 to 93% in 2019 [19,21].
In Q-ECT, patients' self-assessment of HRQoL is obtained using EQ-5D-3L paper questionnaires before ECT, within one week after ECT, and at a six-month follow-up.The proportion of patients who respond to EQ-5D after ECT has increased over time (i.e. from 16% in 2012 to 50% in 2019) [21].

ECT technique
In clinical practice, ECT is usually administered three times per week and using the bidirectional constant-current brief-pulse Mecta (Mecta Corp) or Thymatron (Somatics Inc) devices.Age-based dosing is standard in Sweden (approximately 5mC per year), with some adjustments for sex, anesthesia, concomitant pharmacotherapy, severity of symptoms, and charges required in previous treatment series.
All study participants had unilateral electrode placement (d'Elia) at the first ECT session.The median stimulus at first ECT was: pulse width, 0.5 ms; current, 800 mA; duration, 7 s; frequency, 60 Hz; charge, 307 mC; and the median duration of seizure, 45 s.The pulse width ranged between 0.25-1.00.Only index treatments were included in this study.An index treatment series is the initial treatment, where patients typically receive ECT sessions three times a week either until remission or until the physician judge that the maximum treatment benefit is reached.Therefore, patients in the study sample may have received continuation ECT after the index treatment to prevent the recurrence of depressive symptoms.

Selection of study population
The Swedish version of the International Classification of Diseases 10 th revision (ICD-10) codes were used to identify treatment series with a specified indication.All patients who met the following inclusion criteria were included in the study: age ≥18 years; major unipolar depression (ICD-10 codes F32.1-F32.3,F33.1-F33.3)or bipolar depression (ICD-10 F31.3-F31.5)as indication for treatment; received index treatment and unilateral placement at first ECT session; at least two EQ-5D measurements (either EQ VAS or EQ-5D index at baseline and within one week after ECT); and information on pulse width at first ECT session.If a patient had several index treatment series, only the earliest was chosen.The treatment series included were conducted between 2011 and 2019.The data obtained covered all 21 regions in Sweden, yet more than half of the study population were treated in the three largest regions: Stockholm (38.7%),Västra Götaland (9.1%), and Skåne (9.0%).

Variables
The main explanatory variable was information on pulse width at the first ECT session, categorized into three subgroups: <0.5 ms; =0.5 ms; >0.5 ms.The study designed aimed to follow an 'intention-to-treat' approach, i.e. in this case to classify patients based on their assigned treatment at first ECT session.
The outcome variables were HRQoL (as assessed by EQ-5D-3L index values and EQ VAS) within one week after ECT (primary outcome) and six months after ECT (secondary outcome).HRQoL was selected as outcome measure due to its potential to integrate both benefits and adverse effects into one measure, and to identify the optimal balance between these effects [17].EQ-5D-3L is a generic patient-reported outcome measure on which the respondent rates his/her health today.EQ-5D-3L consists of a descriptive system covering five health dimensions (mobility; self-care; usual activities; pain/discomfort; anxiety/depression, each with three severity levels: no; moderate; extreme problems), and a visual analogue scale (EQ VAS) ranging from 0 (worst) to 100 (best imaginable health state).The descriptive system can identify 243 health states.Each health state can be converted into an EQ-5D index value, representing preferences for different health states.The stated preferences are commonly elicited from the general population in health state valuation studies.EQ-5D index values were calculated using the uK EQ-5D-3L value set [22].
Demographic and clinical characteristics were obtained to examine between-group differences (Table 1) and to assess the generalizability of study findings by comparing the characteristics of the study population to the registry population (Supplementary materials, Table S1).Information about the highest educational level was retrieved from the same calendar year as when ECT was received.The latest available information (2017) was used to describe the highest educational level for patients who received ECT between 2017 and 2019.

Statistical analysis
Demographic and clinical characteristics were presented for the total sample and by subgroups, using frequency (%), mean (standard deviation [SD]) or median (interquartile range [IQR]).Between-group differences were tested using chi-square test, chi-square test for trend, or analysis of variance (ANOVA).
Differences in responses before and after ECT were analyzed in several steps.Individual-level changes in responses to the EQ-5D descriptive system was analysed using the Paretian Classification of Health Change (PCHC) [23].For the total sample, the mean of the differences in EQ-5D index and EQ VAS were examined by paired t-tests and by estimating effect sizes (Cohen's d).A change of 0.082 in the EQ-5D index was considered a minimally important difference (MID) [24].A MID of 7.0 was used as guidance for interpreting EQ VAS scores [25].Effect sizes (Cohen's d) were interpreted using the following thresholds [26]: very small (0.01), small (0.2), medium (0.5), large (0.8), very large (1.2).
In the analysis of the association between pulse width and HRQoL, the outcomes were examined as continuous variables and with adjustment for baseline HRQoL [27].Due to heteroscedasticity (tested using the Bruesch-Pagan test), ordinary least squares (OLS) regression with robust standard errors was used [28].Covariates were included in the regression models based on their expected association with pulse width assignment and treatment effect.Pulse width and HRQoL (EQ-5D-3L index or EQ VAS) at baseline were included in all models (model 1-3).Variables for age, sex, and indication were added in model 2, due to their clinical relevance for the choice of pulse width.In model 3, variables for concurrent medications during ECT (i.e.antidepressants, antipsychotics, benzodiazepines, lithium, antiepileptics) were added due to the possible influence on the seizure, which could also influence the choice of pulse width.Multicollinearity was examined for all covariates in model 3 (i.e. the variance inflation factor [VIF, range 1.012-1.844]did not indicate problems with multicollinearity).
Statistical tests were conducted with a significance level of 0.05.Due to multiple testing of the final model, a Bonferroni correction was applied before interpreting the results of the main analyses (model 3, significance level 0.0125).Statistical analyses were conducted using IBM SPSS Statistics 27.0 (Armonk, NY: IBM Corp) and STATA Statistical Software 15 (College Station, TX: StataCorp LCC).
In addition to the analyses specified in the study protocol [17], interactions were explored.Statistical interactions between pulse width and sex, age, and psychotic features were all tested in the regression models; results are not reported due to lack of statistical significance for these interactions.

Ethical considerations
The study was approved by the regional ethical vetting board in uppsala, Sweden (registration no.2014/174, 2014/174/1, 2014/174/2, 2020-05154).No written consent was required for this study.Before registration in Q-ECT, patients are informed that data may be used for research purposes.Registration is voluntary, with the possibility to opt-out (meaning that the patient has the right to decline registration and to be removed from the register at any time).

Results
Among the 20,052 patients registered in Q-ECT, 5,046 patients were included in the study sample (Figure 1).The study sample and the registry population were similar with regard to age and sex (Supplementary materials, Table S1).Regarding the indication for treatment, the study sample had a slightly larger proportion of unipolar depression (83% vs 82%) and a slightly smaller proportion of psychotic features (17% vs 20%).In addition, the use of 0.5 ms pulse width at first ECT was more frequent in the study sample (72%) than in the registry population (66%).Consequently, the study sample had a smaller proportion of patients receiving <0.5 ms pulse (15% vs 19%) and >0.5 ms pulse (13% vs 15%).
EQ-5D index values before and one week after ECT were available for 4,990 patients and EQ VAS scores for 4,914 patients (4,858 patients completed both) (Table 2).For EQ-5D index and EQ VAS, respectively, the mean of the difference before and one week after ECT was 0.40 (p < .0001,t = 84, df = 5) and 34 (p < .0001,t = 91, df = 5).Effect sizes were large (EQ-5D index, Cohen's d = 1.19) to very large (EQ VAS, Cohen's d = 1.30).By subgroup, the mean difference before and after ECT ranged between 0.38 (<0.5 ms pulse) and 0.40 (0.5 ms pulse) for the EQ-5D index values and between 33 (<0.5 ms pulse) and 35 (>0.5 ms) for EQ VAS scores.On average, patients receiving >0.5 ms pulse had the highest HRQoL before ECT (index 0.33; EQ VAS 27) and after ECT (index 0.73; EQ VAS 62).using the EQ-5D descriptive system, the proportion who reported an improvement ranged from 71% to 73% and a worsening from 5% to 7% (Figure 2a).
A subsample had also completed EQ-5D index (n = 730) and EQ VAS (n = 851) six months after ECT (Table 3).For this subsample, the mean EQ-5D index before ECT was 0.32, and the equivalent for EQ VAS was 27.At six months, the mean EQ-5D index was 0.64 and the mean EQ VAS was 61.On average, patients receiving >0.5 ms pulse had the highest HRQoL before ECT (index 0.36; EQ VAS 31), after ECT (index 0.75; EQ VAS 65), and six months after ECT (index 0.70, EQ VAS 63).The mean difference before ECT and six months after ECT ranged between 0.31 and 0.34 for EQ-5D index (for subgroups 0.5 ms and >0.5 ms pulse, respectively) and between 32 and 35 for EQ VAS (for subgroups <0.5 ms and 0.5 ms pulse, respectively).Compared to before ECT, 67% of patients receiving >0.5 ms pulse at the first ECT session had an improvement on the EQ-5D descriptive system (Figure 2b).The equivalent proportions for the subgroups receiving 0.5 ms and <0.5 ms pulse were 59% and 61%, respectively.There were no statistically significant associations between pulse width and patients' HRQoL measured within one week after ECT, after adjusting for demographic and clinical characteristics (Table 4, Model 3; Supplementary materials, Table S2).Compared to the reference subgroup receiving <0.5 ms, EQ-5D index values six months after ECT were −0.089 lower for patients receiving 0.5 ms, after adjusting for EQ-5D index at baseline, sex, age, depression diagnosis, psychotic features, and concurrent medication (p = .011)(Table 5, Model 3; Supplementary materials, Table S3).The corresponding analysis for examining the association between pulse width and EQ VAS at six months follow-up after ECT showed non-significant results (p = .390).

Discussion
This registry-based study contributes to the understanding of real-world outcomes for patients treated with ECT for unipolar or bipolar depression, specifically focusing on the association between pulse width and HRQoL.All three pulse width subgroups (<0.5 ms; 0.5 ms; >0.5 ms) showed relatively large mean improvements in HRQoL one week after ECT, but no consistent associations were observed between pulse width and HRQoL after ECT when adjusting for HRQoL at baseline, age, sex, indication, and concurrent medication.In the subsample of patients with an EQ-5D index value recorded at  six-month follow-up, the subgroup who received pulse width 0.5 ms had slightly lower mean HRQoL than those receiving <0.5 ms.Nevertheless, these results should be interpreted with caution as they only concerned a subsample of patients and were not confirmed in the corresponding analysis for EQ VAS.Furthermore, the interpretation of outcomes at six months is more sensitive to the possible influence of other events or treatments during the follow-up period that may impact patients' HRQoL, which were not accounted for in this study.
Several other findings related to the outcomes in HRQoL after ECT ought to be highlighted.Independent of the pulse width, more than 70% of the patients in the sample of the current study reported improvements in at least one of the dimensions covered by the EQ-5D descriptive system (i.e.mobility, self-care, usual activities, pain/discomfort, anxiety/ depression) within one week after ECT.Six percent experienced worsening in at least one of the dimensions after ECT, and the remaining 20% experienced no change or mixed change.The average improvement for the total sample was Table 3. Mean EQ-5d index values and mean EQ Vas scores before Ect, within one week after Ect, and six months after Ect (secondary outcomes) for the subsample of patients who completed all three measurements, by subgroups receiving different pulse width at the first Ect session.0.40 for the EQ-5D index and 34 for the EQ VAS.These results were well above the minimally important differences of 0.082 for index values and 7.0 for EQ VAS scores.Still, in our study, most between-group differences for the subgroups receiving different pulse width were below these thresholds.Based on aggregated data for the subsample with a follow-up assessment, the mean improvement observed one week after ECT was maintained at six months.
Assessing similarities or contradictions with previous research is constrained by the limited number of studies on this topic.Our main findings are in line with the results of a previous observational study by Galvez and colleagues [13].By comparing patient subsamples receiving ultra-brief (0.25-0.3 ms) and brief (0.5-1 ms) pulse widths, they found no significant relationship between pulse width and HRQoL in patients treated with unilateral ECT.Compared to the previous study, an important contribution of our study is that the detailed information available from clinical registries enabled adjustments for relevant clinical characteristics for a relatively large sample, e.g.HRQoL at baseline, psychotic features, and use of concurrent medication.Furthermore, the results provide an important contribution to the literature on outcomes after ECT by presenting results for patients receiving the current Swedish standard of 0.5 ms separately from ultra-brief and broader pulse widths.
In the current study, we intended to use EQ-5D as an overall measure of HRQoL that could capture the influence of both benefits and adverse effects into one measure.It should be noted that there may still be benefits to certain pulse width assignments with regard to other outcomes that were not specifically examined within the scope of our study, e.g.speed of treatment response.Previous reviews of the scientific literature highlight the limited evidence available on the topic and present different conclusions regarding the potential advantages in terms of efficacy and/or adverse effects [7,11,12].

Strengths and limitations
The observational study design is associated with both strengths and limitations.A strength of the study is the use of national registry data, representing real-world outcomes for a relatively large population of patients who have been treated with ECT for unipolar or bipolar depression.The detailed information on demographic and clinical characteristics available from the registry enabled adjustments of possible confounding variables in the regression analyses.Another strength of the study was the use of a prespecified study protocol, which contributes to increased transparency of the study methods and reduces the risk of publication bias.However, it should also be noted that real-world data is associated with certain challenges related to study design.Importantly, in clinical practice, the pulse width may have been adjusted during the treatment series.The pulse width at first ECT session was used for classifying patients to reflect the intention to treat, and therefore, no adjustments were made in the analysis.
By design, the use of observational data limits the interpretation of causal relationships between treatment and outcomes.Further research is still needed to explore potential causal relationships, possibly by exploring other study designs or statistical methods to mitigate the impact of confounding by indication.In addition, other treatment modalities may impact HRQoL outcomes.For example, for some algorithms used (including the commonly used Thymatron '0.5 ms double dose' setting), pulse widths increase with increasing charge and longer pulse widths are therefore also associated with higher charges.Importantly, this study addresses associations and not causal relationships between pulse width and HRQoL.
Another study limitation was that only a subsample of the total registry population (38%) had responded to EQ-5D before and after ECT and were included in this study.The comparison of basic demographic and clinical characteristics for patients in the study sample and the registry population indicated similar distributions regarding demographic and clinical characteristics such as age, sex, and initial treatment setting.Yet, slight differences between the samples were observed regarding some clinical characteristics, e.g.pulse width and number of ECT sessions.
In addition, the study sample was reduced further in the analyses.First, a restraint of our final regression model was the reduced sample size following the inclusion of  concurrent medication as a covariate, i.e. 39% of the study sample had incomplete data concerning the use of one or several concurrent medications.Second, the results from the follow-up at six months after ECT should be interpreted with caution due to the substantial loss to follow-up.For those included in the study samples, the response rates at six-month follow-up were 12% for EQ-5D index values and 14% for EQ VAS scores.

Implications and recommendations
The results from the current observational study did not indicate any advantages of using one pulse width over others in terms of the overall impact on patients' HRQoL after ECT.The interpretation of HRQoL outcomes at six months after ECT is less certain due to the large proportion of loss to follow-up.Thus, further research is needed to better understand the long-term outcomes, and their predictors, using a longer follow-up time after ECT.

Conclusion
On average, patients with unipolar or bipolar depression had significant HRQoL improvements after ECT, independent of the pulse width received at the first ECT session.In conclusion, the current study showed no robust associations between pulse width and HRQoL either within one week or at six-months follow-up after ECT.Further explorations are encouraged to contribute to the understanding of how to optimize the treatment effect in patients receiving ECT for unipolar or bipolar depression.

Figure 1 .
Figure 1.flow chart describing the selection of study sample.

Table 1 .
Baseline demographic and clinical characteristics of the total sample and by subgroup defined by pulse width (in milliseconds [ms]) at first Ect session.subgroups(pulse width at first Ect session) a Pearson chi-square test.b analysis of variance (aNoVa).c chi-square test for trend (linear-by-linear association).

Table 2 .
Mean EQ-5d index values and mean EQ Vas scores before Ect and within one week after Ect (primary outcomes) for all patients in the study sample (n = 4,990), by subgroups receiving different pulse width at first Ect session.

Table 4 .
results from ordinary least squares (ols) regression with robust standard error, examining the association between pulse width (expressed in milliseconds [ms]) and health-related quality of life within one week after Ect (primary outcomes), with adjustment for potential confounding variables.

Table 5 .
results from ordinary least squares (ols) regression with robust standard error, examining the association between pulse width and health-related quality of life six months after Ect (secondary outcomes), with adjustment for potential confounding variables.EQ-5d index -six months after Ect EQ Vas -six months after Ect