If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Use of electroconvulsive therapy for individuals receiving inpatient psychiatric care on a nationwide scale in France: Variations linked to health care supply
EPS Barthélémy Durand, Psychiatry Department, Avenue Du 8 Mai 1945, 91150, Etampes, FranceInstitut de Recherche et Documentation en économie de la santé (IRDES), 117 bis Rue Manin, 75019, Paris, France
A comprehensive understanding of variations in the use of ECT among health care providers is currently lacking.
•
Availability of routinely collected data on ECT in France enables the development of innovative research on this topic.
•
A conceptual framework based on the international literature on medical practice variations was used.
•
This research demonstrates some consistency between actual ECT use and national clinical guidelines.
•
Variations in ECT use were linked to health care supply characteristics, questioning access to quality mental care.
Abstract
Background
A comprehensive understanding of variations in the use of electroconvulsive therapy (ECT) among health care providers in charge of ECT referrals is lacking.
Objective
Our objectives were to document ECT use and its variations on a nationwide scale in France and to identify the factors that were significantly associated with these variations.
Methods
Administrative health claims data on hospitalization were used to perform a descriptive analysis of ECT use for adult patients receiving inpatient psychiatric care in mainland France in 2019 and its variations across hospitals in charge of ECT referrals. Based on a conceptual framework drawn from the literature on medical practice variations, a multilevel logistic regression was then conducted to identify patients, hospitals and contextual characteristics that were significantly associated with ECT treatment using non-ECT-treated patients receiving inpatient psychiatric care as the reference population.
Results
Patients receiving ECT (n = 3288) were older, more frequently female and had more severe diagnoses than other patients seen in inpatient care (n = 295,678). Significant variations were observed in the rate of ECT use across hospitals (n = 468), with a coefficient of variation largely above one. In the multivariable analysis, ECT treatment was associated with patient characteristics (which accounted for 6% of the variations) but also with characteristics of the hospitals and their environments (44% of the variations), including the type of hospital and its distance to the closest facility providing ECT.
Conclusions
Variations in ECT use were strongly linked to health care supply characteristics, which raises questions about access to quality mental health care.
Numerous studies have demonstrated the efficacy and safety of electroconvulsive therapy (ECT) in severe and pharmacoresistant depression, mood disorders and schizophrenia [
Risk of serious medical events in patients with depression treated with electroconvulsive therapy: a propensity score-matched, retrospective cohort study.
The role of electroconvulsive therapy (ECT) in bipolar disorder: effectiveness in 522 patients with bipolar depression, mixed-state, mania and catatonic features.
]. Regardless of this strong evidence base, significant variations in the use of this procedure among geographical areas or health care providers have been described in several countries [
]. Several explanatory hypotheses can be drawn from the literature on unwarranted medical practice variations, which represent the gap between practice and evidence-based medicine. One of these hypotheses is the difference in preferences or habits of health care providers, in particular in a context of uncertainty [
]. For ECT, this can notably be linked to the high efficacy of the procedure, as opposed to the negative representations among some health professionals [
], and to the continuum of clinical manifestations of mental disorders, which can make it difficult to determine the ideal timing of the procedure among a sequence of alternatives [
]. Other hypotheses include trade-offs between opportunities and constraints of health care providers that are linked to their working environment (for instance, the availability of equipment such as technical platforms for ECT) [
]. One approach that can be adopted relies on identifying associated factors, for which it is easier to determine whether they should legitimately influence medical decisions [
]. These factors can then be targeted by policies aimed at reducing unwarranted variations. In the international literature, factors that were found to be associated with differences in ECT use not only include patient characteristics, such as age, sex and diagnostic group, but also health insurance status or neighborhood socioeconomic deprivation, as well as characteristics of health care providers and their environments, such as academic status and geographic isolation [
The current evidence base on variations in ECT use offers avenues for further exploration, as existing studies are scant in many countries and present limitations. Some previous research has been carried out on small scales, considering only a few health care providers or a single region [
], which leads to questions regarding the comprehensiveness of the information collected. Furthermore, research has often focused on variations among health care providers conducting ECT [
] rather than on those prescribing the procedure. The latter do not always have in-house technical resources to conduct ECT themselves but represent a more interesting focus of analysis, as they are in charge of the decision for an ECT referral. A better understanding of medical practice variations requires a focus on the decisional rather than on the operational level. Finally, existing studies have often considered a limited number of potential factors that are associated with variations in ECT use, in particular regarding health care supply characteristics, and the associations were often investigated in a rather descriptive manner, limiting their interpretation.
In this context, our objectives were to document ECT use and its variations, specifically among health care providers prescribing the procedure, as well as to identify the factors significantly associated with these variations on a nationwide scale in France using recent administrative health claims data.
2. Materials and methods
2.1 Setting
In France, public psychiatry has been organized by catchment area since the 1960s, with the aim of providing equitable and free mental care in the territory under the supervision of hospital teams [
]. In line with this historical organization, hospitals, whether public or private, still play a determining role in the coordination of care for severe and persistent mental disorders. They are often in charge of following patients with such disorders and organizing their inpatient or outpatient care [
]. Thus, while ECT is only performed in hospitals with a dedicated technical platform – sometimes without a psychiatric department – the decision to resort to this procedure relies heavily on the hospital in charge of patients’ psychiatric follow-up. France is a country where ECT has been part of the available therapeutic options since its discovery [
France has a nationwide system for the collection of standardized administrative and medical data on care provided in hospital settings, either for somatic care (Programme de médicalisation des systèmes d'information, PMSI-MCO) [
]. General hospitals have reported their ECT activity in the PMSI-MCO database for years, as this was linked to funding for the procedure in the framework of an activity-based payment system. This has not been the case for hospitals that specialize in psychiatry, which are mainly funded by a global allocation that is not directly related to their activity. A national health claims registry of ECT procedures, coded not only by general hospitals, but also by psychiatric hospitals, was set up in 2017 by the technical agency for information on hospital care (Agence technique de l'information surl'hospitalisation, ATIH), which is in charge of managing administrative health claims data on hospitalizations. This registry has been available since then in the PMSI-MCO and RIM-P databases, which were our main data sources. They include information on inpatient, part-time and ambulatory care provided in hospital settings at the individual patient level, as well as on patients who are treated and on the hospitals providing care. Permanent access to these data is granted to the lead institution involved in this research, which does not require any specific ethical approval or informed consent for accessing these data, which are fully anonymized [
Complementary databases providing publicly available information at the aggregated hospital or geographical level were used more sporadically to obtain additional data on the characteristics of the hospitals providing care (annual surveys or directories of health care providers) [
Our study population included all patients aged 18 or over who received psychiatric inpatient care in mainland France in 2019 (year with the most complete data since the launch of the national health claims registry of ECT and latest year available before the COVID-19 pandemic). We chose to focus on a relatively homogeneous population for whom indications for ECT are the most frequent. Thus, we excluded patients who were seen only for part-time hospitalizations – the organization of this form of activity being very heterogeneous among hospitals – and ambulatory care, which rarely leads to a referral for ECT treatment in France (more than 90% of patients receiving ECT in a given year had psychiatric inpatient care the same year – authors’ own calculation using the PMSI-MCO and RIM-P databases). We also excluded patients from hospitals that are dedicated to children patients (but sometimes also follow a few young adults) for whom care modalities are very specific, particularly concerning the collection of consent, and patients hospitalized in overseas territories where the organization of psychiatry differs from mainland France and where the collection of hospital activity data is less comprehensive. We did not carry out any further exclusion based on diagnostic groups, as the classification systems used in health claims data (here, the International Classification of Diseases, Tenth Revision – ICD-10 [
], and the occurrence of at least one inpatient stay was already a sign of the severity of the mental disorder.
2.4 Units of analysis
We focused on variations in ECT use among health care providers rather than among geographical areas, which can tend to mask differences among providers located in the same territory. Health care providers considered in our research were those in charge of the ECT referral (decision occurring in hospital settings), which are not always those in charge of performing it (as this is linked to the availability of technical platforms). Information on health care providers in the national administrative health claims data used for this research only includes the hospital level, and not the individual physician level within the hospitals. We therefore focused on hospitals in charge of the psychiatric follow-up of our study population, defined as the hospital where the included patients spent the highest number of inpatient days between the start of the year 2018 and the first ECT procedure of 2019, or the end of the year 2019 for those who did not undergo ECT.
2.5 Variable of interest
Our binary variable of interest was, for all patients included, whether they had received ECT at least once during the year 2019. This variable was therefore calculated at the individual patient level. We also computed the rate of ECT use in each hospital in charge of psychiatric follow-up by dividing the number of patients they followed who were seen in inpatient care and received ECT in 2019 by the total number of patients they followed who were seen in inpatient care in 2019.
2.6 Factors potentially associated with ECT use
As medical practice is a complex decision-making process, we used a conceptual framework based on previous research [
Utilisation d’un modéle de régression logistique à deux niveaux dans l’analyse des variations de pratique médicale : à propos de la césariennne prophylactique.
] that grouped potential factors that are associated with practice variations into three categories: patient characteristics (the demand side), health care provider characteristics (the supply side) and characteristics of the environment (practice context). Among these categories, several factors drawn from the literature [
], or clinicians’ hypotheses, should particularly be considered for the study of variations in ECT use (Table 1).
Table 1Factors potentially associated with the use of ECT, main hypothesis of association and construction of variables.
Potential factor associated with the use of ECT
Main hypothesis of association
Construction of the variable
Patient demographics
Age
Patient demographics could be correlated to their health needs (representing for instance severity markers) and illustrate differences in clinical needs among patients leading to variations in ECT use. Age, in particular, could be associated with situations of drug resistance and multiple therapeutic attempts over time.
Extracted from the 2019 RIM-P database, age or gender during the first inpatient stay (or the following stay if missing information).
Gender
Patient socioeconomic characteristics
Inclusion (or not) in the scheme covering health care costs for low-income groups (‘couverture maladie universelle complémentaire’, CMU-C)
Patient individual socioeconomic deprivation level could be linked to differing health needs and different care pathways within the mental health system, resulting in differential access to ECT.
Extracted from the 2019 RIM-P database, at least one contact with the mental health system associated with a coverage through the CMU-C scheme.
Neighborhood socioeconomic deprivation index (FDep) at the zip code of residence and corresponding quintile
Socioeconomic characteristics of patients' living environment could be linked to differing health needs and different care pathways within the mental health system, resulting in differential access to ECT.
Constructed mainly using 2015 census data (latest year available at the time of the research) taking into account the median household income, the percentage of high school graduates in the population aged over 15 years of age, the percentage of blue-collar workers in the active population and the unemployment rate [
]. This variable was then coded in quintiles (the higher the quintile, the higher the deprivation).
Patient clinical characteristics
Diagnostic group
Patients' diagnoses could reflect different clinical presentations, which could lead to variations in the use of ECT to meet different health needs in line with national clinical guidelines [
Extracted from the 2019 RIM-P and PMSI-MCO databases, considering primary and secondary psychiatric diagnoses over the whole year for all included patients' inpatient stays to build the seven following exclusive diagnostic groups: 1/mild depression: only diagnoses of depression without any severity sign and no diagnosis of other mood or psychotic disorder (ICD-10 codes: F320/F321/F330/F331 – F2/F30/F31/F061/F202 excluded); 2/severe depression: at least a diagnosis of depression with severity signs but no diagnosis of bipolar or psychotic disorder (ICD-10 codes: F32/F33 – F30/F31/F2/F320/F321/F330/F331/F323/F333/F061/F202 excluded); 3/depression with psychotic or catatonic characteristics: at least a diagnosis of depression with psychotic or catatonic characteristics but no diagnosis of bipolar disorder or psychotic disorder (ICD-10 codes: F323/F333/F061/F202 – F30/F31/F2 excluded); 4/bipolar depression: at least a diagnosis of depression associated to a diagnosis of bipolar disorder without manic, hypomanic or mixed episode (ICD-codes: F32/F33 and F30/F31 – F310/F311/F312/F316 excluded); 5/bipolar disorder with manic, hypomanic or mixed episode (ICD-codes: F30/F310/F311/F312/F316); 6/psychotic disorder without mood disorder (ICD-10 codes: F2 – F30/F31/F32/F33/F061/F202 excluded); 6/other diagnoses (all other diagnoses leading to a psychiatric inpatient stay, including in particular anxiety and substance abuse disorders; ICD-10 codes: F – excluded z/F2/F30/F31/F32/F33/F061/F202/).
A history of therapeutic seclusion
These variables are markers of the severity of patients' mental disorders which could lead to variations in the use of ECT to address different health needs in line with national clinical guidelines [
Extracted from the 2018 and 2019 RIM-P database, at least one occurrence of therapeutic seclusion in 2018 or 2019 (until the first ECT procedure of 2019).
A history of suicide attempt
Extracted from the 2018 and 2019 PMSI-MCO database, at least one suicide attempt followed by a hospitalization for somatic care in 2018 or 2019 (until the first ECT procedure of 2019) identified by at least one secondary diagnosis with an ICD-10 code in X60 to X84 [
Utilisation d’un modéle de régression logistique à deux niveaux dans l’analyse des variations de pratique médicale : à propos de la césariennne prophylactique.
Characteristics of the hospital in charge of the psychiatric follow-up
Type of hospital
The use of ECT could vary according to the institutional characteristics of the hospital in charge of the psychiatric follow-up, for instance because of differing schools of thought.
Extracted from annual surveys or directories of health care providers for the year 2019 and construction of five exclusive categories: public hospital with teaching activities, pluridisciplinary public hospital without teaching activities, public hospital specialized in psychiatry without teaching activities, private-non-profit hospital, private for-profit hospital.
Contextual characteristics
Distance between the hospital in charge of the psychiatric follow-up and the closest facility providing ECT (availability of a technical platform)
A low distance between the hospital in charge of the psychiatric follow-up and the closest facility providing ECT could increase access to this procedure (better knowledge of the practice and its practicality, easier exchanges between teams …).
Computed using the Droma distance calculator for the year 2018 considering the distance in kilometers by road when traffic is running smoothly [
Using clinical practice variations as a method for commissioners and clinicians to identify and prioritise opportunities for disinvestment in health care: a cross-sectional study, systematic reviews and qualitative study. 3. Health Services and Delivery Research,
2015
Density of anesthetists in the local county (‘département’) of the hospital in charge of the psychiatric follow-up
A high density of anesthetists in the territory in which the hospital of the psychiatric follow-up operates could be associated to an increased availability of ECT procedures (more possibilities to open technical platforms).
Extracted from the 2020 national directory on available health care professionals, including anesthetists who were fully salaried in hospitals and those who were both salaried in hospital and self-employed, and considered for 100,000 inhabitants.
The main clinical and sociodemographic characteristics of the study population were described by numbers and percentages, according to whether patients received ECT.
Variations in ECT use were described by comparing the rate of ECT use across each hospital in charge of psychiatric follow-up and in comparison to the average national rate [
Using clinical practice variations as a method for commissioners and clinicians to identify and prioritise opportunities for disinvestment in health care: a cross-sectional study, systematic reviews and qualitative study. 3. Health Services and Delivery Research,
2015
]. We therefore calculated the national mean, standard deviation (SD), median, interquartile range and coefficient of variation (CV) of the rate of ECT use.
To identify patients, hospitals and contextual characteristics significantly associated with ECT use, all other things being equal, a logistic regression was carried out with non-ECT-treated patients receiving inpatient psychiatric care as the reference population. To account for the nested structure of the data of patients followed in the same hospital, we conducted a hierarchical model with two levels: the patient level and the hospital level. First, we carried out an empty model (Model 1) that included only a random intercept at the health care provider level and no explanatory variables to confirm the existence of a center effect. Second, we introduced individual patient characteristics in the model (Model 2). Third, we added variables calculated at the health care provider level, i.e., characteristics of the hospitals in charge of psychiatric follow-up and of the territory in which they operate (Model 3, see equation in Appendix A). All patient, health care provider and environmental characteristics that were identified in the literature and in clinicians’ hypotheses as associated with ECT use (Table 1) were introduced as explanatory variables after controlling for collinearity and carrying out preliminary multilevel univariate analyses. To determine the strength and direction of the association between the probability of using ECT and each explanatory variable in the final model, we estimated the odds ratios (ORs) and their 95% confidence intervals (95% CIs), for which interpretation is similar to risk ratios, due to the rarity of ECT use [
]. A statistical significance level of 5.0% was consistently used. For each model, we also calculated the intraclass correlation coefficient (ICC), which is the proportion of variance that is accounted for by the hospital level [
To test the robustness of the final model, two sensitivity analyses were conducted. The first aimed to verify whether factors associated with ECT use would have been different if we had only focused on the initiation of treatment. This, therefore, consisted of modifying the binary variable of interest to only consider, among patients who received ECT in 2019, those who did not receive ECT in the previous years for which data were available, i.e., 2017 and 2018. The second sensitivity analysis aimed to determine whether factors associated with ECT use would have been different if we had focused only on patients with diagnoses for which ECT use is the most recommended based on national clinical guidelines (in particular for severe or complex depression as well as bipolar and psychotic disorders) [
]. We therefore excluded all patients in the mild depression diagnostic group or with diagnoses included in the ‘other’ diagnoses category (see Table 1).
2.9 Software
Data cleaning, management and analyses were carried out with SAS Enterprise guide version 8.3 and R version 4.0.2.
3. Results
3.1 Study population
All in all, based on the French national health claims registry of ECT procedures, and before applying our inclusion and exclusion criteria, it is estimated that 3705 individuals received ECT in 2019, corresponding to a rate of 0.6 per 10,000 inhabitants-year.
Among the 419,794 patients hospitalized in a psychiatry department in France in 2019, 298,966 were patients aged over 18 who received inpatient care in mainland France and met our inclusion and exclusion criteria (Fig. 1). A total of 3288 (1.1%) patients had received ECT in 2019.
Fig. 1Flowchart for the selection of the study population.
Patients who received ECT were more often female, older, living in less deprived areas, suffering from severe or complex depression, presenting with severity markers and were less often included in the scheme covering health care costs for low-income groups than other patients receiving inpatient care (Table 2).
Table 2Characteristics of the study population.
Characteristic
Number (%)
Patients with ECT (n = 3288)
Patients without ECT (n = 295,678)
Demographics
Female
2099 (63.84)
147,559 (49,91)
Age (in years)
Between 18 and 30
223 (6.78)
57,178 (19.34)
Between 31 and 40
325 (9.88)
52,397 (17.72)
Between 41 and 50
514 (15.63)
62,152 (21.02)
Between 51 and 60
703 (21.38)
58,701 (19.85)
Between 61 and 70
762 (23.18)
36,056 (12.19)
71 or older
761 (23.14)
29,194 (9.87)
Socioeconomic characteristics
Inclusion in the scheme covering health care costs for low-income groups (CMU-C)
171 (5.20)
37,214 (12.59)
Neighborhood socioeconomic deprivation index (FDep)
The included patients were seen in 468 hospitals in charge of their psychiatric follow-up. The average rate of ECT use in these hospitals amounted to 1.1 per 100 patients seen in psychiatric inpatient care, with 137 hospitals (29.3%) for which no patient received ECT in 2019. The coefficient of variation was far superior to one (2.8), indicating strong variability across the hospitals (Table 3), which remained even when removing hospitals with no ECT referrals in 2019 (CV = 2.3).
Table 3Rate of ECT use across hospitals in charge of the psychiatric follow-up.
Mean (SD)
Median (interquartile range)
Range
CV
Rate of ECT use for 100 patients receiving psychiatric inpatient care
There were statistically significant variations among the hospitals in charge of psychiatric follow-up in the empty model, with a variance significantly different from zero (p < 0.0001), which confirmed the need to run a random-intercept model. In addition, the MOR was distinctly superior to one, confirming the significance of the center effect. A large part (69%) of the total variation in the rate of ECT use was related to differences within the hospitals in charge of psychiatric follow-up (intrahospital variations), while 31% of the total variation resulted from differences among hospitals (interhospital variation).
Explanatory variables introduced at level 1 (individual patient characteristics) explained 6% of the variations among the hospitals (Model 2), while explanatory variables introduced at level 2 (characteristics of the hospitals and their context) explained nearly 45% of these variations (Model 3) [Table 4].
Table 4Estimation of random effects in the hierarchical model.
Model 1 (null model with no explanatory variables)
Model 2 (with individual patient characteristics)
Model 3 (with individual patients characteristics and characteristics of hospitals and of the territory in which they operate)
The results of the final model (Model 3) confirmed the association of ECT use with individual patient characteristics but also, and more significantly, with characteristics of the hospital in charge of psychiatric follow-up and of the territory in which it operates (Table 5).
Table 5Factors associated with the use of ECT.
Univariate analysis
Multivariable analysis
Characteristic
OR
95%CI
p-value
Adjusted OR
95%CI
Adjusted p-value
Patient characteristics
Female gender (ref = male gender)
1.56
1.45–1.68
<0.0001
1.16
1.08–1.26
0.0001
Age (in years) (ref = 18–30)
Between 31 and 40
1.67
1.40–1.98
<0.0001
1.54
1.29–1.83
<0.0001
Between 41 and 50
2.26
1.92–2.65
1.92
1.63–2.26
Between 51 and 60
3.22
2.76–3.75
2.49
2.13–2.92
Between 61 and 70
5.74
4.93–6.69
4.04
3.45–4.73
71 or older
6.88
5.89–8.02
4.98
4.25–5.85
Inclusion in the scheme covering health care costs for low-income groups (ref = no)
0.44
0.37–0.51
<0.0001
0.67
0.57–0.79
<0.0001
Neighborhood socioeconomic deprivation index (ref = fifth quintile)
The higher the quintile, the higher the deprivation.
First quintile
1.24
1.08–1.43
0.0111
1.14
0.99–1.32
0.35
Second quintile
1.14
0.99–1.32
1.09
0.94–1.26
Third quintile
1.06
0.92–1.21
1.04
0.91–1.20
Fourth quintile
1.02
0.89–1.17
1.02
0.89–1.17
Diagnostic group (ref = severe depression)
Mild depression
0.33
0.27–0.40
<0.0001
0.36
0.30–0.44
<0.0001
Depression with psychotic or catatonic characteristics
3.27
2.93–3.65
3.24
2.90–3.62
Bipolar depression
2.22
2.01–2.46
2.08
1.88–2.31
Bipolar disorder with manic, hypomanic or mixed episode
0.78
0.67–0.91
0.77
0.65–0.90
Psychotic disorder without mood disorder
0.42
0.37–0.47
0.49
0.43–0.55
Other diagnoses
0.10
0.09–0.12
0.13
0.11–0.15
A history of therapeutic seclusion (ref = no)
2.23
1.95–2.55
<0.0001
2.94
2.55–3.39
<0.0001
A history of suicide attempt (ref = no)
1.49
1.28–1.72
<0.0001
1.47
1.27–1.71
<0.0001
Characteristics of the hospital in charge of the psychiatric follow-up
Type of hospital (ref = private-for-profit hospital)
Public hospital specialized in psychiatry without teaching activities
0.61
0.43–0.85
<0.0001
1.01
0.76–1.34
0.03
Pluridisciplinary public hospital without teaching activities
0.59
0.43–0.81
0.96
0.73–1.28
Public hospital with teaching activities
3.44
2.18–5.43
1.87
1.23–2.85
Private-non-profit hospital
0.58
0.37–0.89
0.87
0.59–1.28
Characteristics of the territory in which the hospital in charge of the psychiatric follow-up operates
Distance between hospital in charge of psychiatric follow-up and closest facility providing ECT in km (ref>48.3 km)
On-site
8.26
6.05–11.26
<0.0001
6.09
4.46–8.33
<0.0001
≤12.7
2.27
1.67–3.10
1.78
1.30–2.44
(12.7; 48.3]
1.33
0.97–1.81
1.21
0.89–1.64
Density of anesthetists partly or fully salaried in hospitals in the local county of the hospital in charge of the psychiatric follow-up per 100,000 inhabitants
1.01
1.00–1.01
<0.0001
1.00
1.00–1.00
0.14
a The higher the quintile, the higher the deprivation.
All patient characteristics included were significantly associated with ECT use in the univariate analysis as well as in the final model, except for the neighborhood socioeconomic deprivation index at the patients' zip codes. Regarding demographics and socioeconomic characteristics, all other things being equal, women had 1.2 higher odds of receiving ECT than men, patients over 70 years old had nearly five times higher odds of receiving ECT than patients under 30 years old, and patients included in the scheme covering health care costs for low-income groups had 1.5 times lower odds of receiving ECT than patients not included in this scheme. Regarding clinical characteristics, patients’ diagnostic groups were significantly associated with ECT use. In particular, patients diagnosed with mild depression had nearly three times lower odds of receiving ECT than patients diagnosed with severe depression, while patients diagnosed with depression with psychotic or catatonic characteristics or bipolar depression had three and two times higher odds of receiving ECT, respectively. A history of therapeutic seclusion and of suicide attempts were also significantly associated with higher odds of receiving ECT (Table 5).
Overall, the type of hospital in charge of psychiatric follow-up was significantly associated with ECT use in the univariate analysis, as well as in the final model. All other things being equal, patients followed up in public hospitals with teaching activities had the highest odds of receiving ECT in comparison to patients followed up in other types of hospitals (Table 5).
Regarding the characteristics of the territory in which the hospital in charge of psychiatric follow-up operates, only the distance to the closest facility providing ECT was significantly associated with the use of ECT in the multivariable analysis. Patients followed up in hospitals with an on-site technical platform for ECT had six times higher odds of receiving ECT than those who were followed up in hospitals which were the furthest from such platforms (Table 5).
3.4 Sensitivity analyses
In the two sensitivity analyses conducted, factors that were significantly associated with ECT use were found to be similar to those of our main analysis (Appendix B), and health care supply characteristics also explained the largest share of the variations.
4. Discussion
Using the most comprehensive data source that is currently available on ECT treatment in France, we found a rate of use of this procedure of 0.6 per 10,000 inhabitants-year. This places France in an intermediate position among the countries for which this rate is available and where it ranges from 0.04 to 5.10 per 10,000 inhabitants-year, but below the international average estimated at 2.34 patients per 10,000 inhabitants-year [
]. When specifically focusing on adult patients receiving psychiatric inpatient care in mainland France, we found an average rate of ECT use of 1.1 per 100 patients. Significant variations in ECT use were observed among hospitals in charge of psychiatric follow-up. Individual patient characteristics linked to health needs (such as age, diagnostic groups or severity markers) explained a lower part of the variations resulting from differences among the hospitals than the characteristics of the hospitals and of the territory in which they operate. Higher odds of receiving ECT were notably found for patients followed in public hospitals with teaching activities and in hospitals closest to a technical platform for ECT.
The rate of ECT use for 100 adult patients seen in psychiatric inpatient care found in our study is consistent with rates in European countries estimated to range from 0.6 to 14 [
]. The significant variations found in ECT use are also not surprising when considering the international literature that underscored the uneven use of this procedure in every national context that has been investigated [
]. However, direct comparison of the extent of these variations is precluded by our focus on variations among health care providers (rather than among geographical areas) in charge of the ECT referral (rather than among those providing ECT), which are more appropriate units of analysis in the frame of research on medical practice variations. Having this concern in mind, our findings appear to be consistent with previous research that showed that factors such as teaching activities or the distance to an available ECT platform could play a stronger role in the decision to resort to ECT than patients’ needs [
The significance of the variations in ECT use in France raises questions in a context where mental health care is organized by catchment area aimed at guaranteeing equal access to care with no diverging legislation on ECT across territories, contrary to what is observed in some federal countries [
]. The fact that these variations were more strongly associated with the characteristics of health care supply than of patients is also of concern for the quality of mental care. Among these characteristics, higher odds of receiving ECT, all other things being equal, were observed for patients who were followed up in hospitals with teaching activities (which were all general hospitals with multiple medical specialties). This finding may be linked to easier access to anesthetists, to a more complex and severe case mix that is not captured by available data and to specific funding mechanisms for ECT provided in general hospitals (based on actual activity assessed through administrative health claims data that supports comprehensive data reporting). This finding could also result from the stronger neurosciences background of academic psychiatrists and a related easier diffusion of national clinical guidelines in hospitals with teaching activities. This supports the need to harmonize this diffusion across all settings so that clinicians with little experience with ECT can more easily recognize cases in which it is an adequate indication. Strong attention to the provision of adequate and homogenous ECT training, during the residency and during the continuous professional training of psychiatrists, would help reduce the differences observed across practice settings. The higher odds of receiving ECT found for patients who were followed up in a hospital that was closest to a technical platform, which had already been suggested by several previous studies [
], can be interpreted in several ways. It could be linked to the existence of supplier-induced demand, which implies that the proximity of a technical platform for ECT would make clinicians more likely to prescribe this treatment, or to inequities of access to ECT for patients whose health status requires it, but who are followed up in a hospital that is far from the necessary technical resources. Overall, it encourages stronger health care planning strategies and, in particular, the development of clearly defined referral pathways from each hospital in charge of psychiatric follow-up toward a technical platform for ECT to guarantee equality in access to the technique among the patients with indications for this treatment.
Despite the concerning associations found with health care supply characteristics, a relative consistency was observed between national clinical guidelines [
] and patient characteristics associated with ECT use. Patients with complex mood disorders and severity markers were the most likely to receive ECT, as were older patients who were more likely to have developed pharmacological resistance to psychotropic drugs over time. Part of the variations found in our study could be hypothesized to be linked to milder mental disorders for which the evidence base on the risk-benefit ratio of ECT is weaker [
]. However, our sensitivity analysis focusing on patients with the most severe and complex disorders also showed a significant association of health care supply characteristics with ECT use. While middle and upper socioeconomic groups were disproportionately represented among ECT recipients in other research [
], definite conclusions were not possible in our study, as a significant association was only observed for inclusion in the scheme covering health care costs for low-income groups, which gives an incomplete picture of individuals struggling with financial hardships.
Our findings should be interpreted considering several limitations. First, there are methodological limitations. We indeed carried out a retrospective analysis of data that does not allow direct conclusions on the causality of the observed associations, which is in line with an exploratory approach in a context where no large-scale and robust data had been available thus far. Additionally, ECT remains a rare event, for which previous researchers have expressed concerns regarding the use of logistic regressions [
]. However, the problem is mostly linked to the possibility of a small number of occurrences for the rarer of the two outcomes. The nationwide scale enabled us to include a high number of ECT procedures, and we used a reasonable number of explicative variables that were mainly used as categorical variables, as recommended for rare events [
Second, there are limits linked to data availability, which may explain why the factors explored did not fully account for all the variations in ECT use. Administrative health claims records indeed include scarce information on patients' socioeconomic and clinical situations, potential coding differences of diagnostic groups according to providers, and no data on patients' and clinicians’ views toward ECT or on the theoretical orientation of hospital departments. Despite these limitations, the availability of nationwide routinely collected data on ECT (not limited to the beneficiaries of specific public or private insurance schemes as in some other countries such as the US) puts France in a favorable position to be the setting for innovative research. Our study considered a wide range of associated factors, and both accounted for the hierarchical structure of the data and focused on providers in charge of the therapeutic decision. Qualitative studies collecting a larger number of patient characteristics, and information at the individual clinician level, would usefully complement our first findings, as would sociological research focusing on the ECT training of young psychiatrists and its impact on their future practice. Perspectives for further research also include quantitative research considering several years of data in a longitudinal approach as well as focusing on the intensity of ECT treatment and its impact on patient pathways.
5. Conclusions
Providing data on ECT use on a large scale through the development of routine monitoring electronic systems [
]. This research conducted in the French context demonstrates some consistency between actual ECT use and national clinical guidelines but underscores the variations linked to health care supply, which raises questions about access to quality mental health care. Our findings support the development of clearly defined referral pathways among hospitals providing psychiatric care and technical platforms for ECT, so that this treatment can be available for all patients who require it.
Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Availability of data and materials
The datasets that were generated and/or analyzed during the current study are available from the corresponding author on request and in compliance with requirements from the French Data Protection Authority.
Ethics approval and consent to participate
Permanent access to the administrative health claims data that were used is granted to the lead institution involved in this research (Irdes), which does not require any specific ethical approval or informed consent for accessing these data, which are fully anonymized. No informed consent was required from the patients, as this research did not involve human and/or animal experimentation and because the analyzed data were entirely anonymized. Other data sources that were used only provided publicly available information.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgments
The authors are indebted to Marie-Odile Safon and Véronique Suhard for their help with the literature review, to Charlène Le Neindre for her help with the Droma distance calculator and to all their Irdes colleagues who provided useful comments on a first version of this research – and, in particular, Julien Mousquès. The authors also thank the participants of the Journées Neurosciences Psychiatrie Neurologie, of the 34thEuropean College of Neuropsychopharmacology Congress and of the 13thFrench Congress of Psychiatry for their valuable comments.
Appendix A and Appendix B. Supplementary data
The following is the Supplementary data to this article:
Risk of serious medical events in patients with depression treated with electroconvulsive therapy: a propensity score-matched, retrospective cohort study.
The role of electroconvulsive therapy (ECT) in bipolar disorder: effectiveness in 522 patients with bipolar depression, mixed-state, mania and catatonic features.
Utilisation d’un modéle de régression logistique à deux niveaux dans l’analyse des variations de pratique médicale : à propos de la césariennne prophylactique.
Using clinical practice variations as a method for commissioners and clinicians to identify and prioritise opportunities for disinvestment in health care: a cross-sectional study, systematic reviews and qualitative study. 3. Health Services and Delivery Research,
2015
☆Meeting presentations: Journées Neurosciences Psychiatrie Neurologie, Paris, France, 1–2 July 2021; 34thEuropean college of neuropsychopharmacology Congress, Lisbon, Portugal, 2–5 October 2021; 13thFrench Congress of Psychiatry, Montpellier, France, 1–4 December 2021.
00842-1&id=gr1.jpg){kind=link}