Characteristics of patients with depression initiating or switching antidepressant treatment: baseline analyses of the PERFORM cohort study

The PERFORM study is a 2-year observational cohort study in outpatients with MDD enrolled by either a primary care physician or a psychiatrist at 194 sites in five European countries (France, Germany, Spain, Sweden and the UK). In each country the study sites were selected to reflect the national proportions of these clinicians treating patients with depression. At enrolment, eligible patients were aged 18–65 years, had a current diagnosis of MDD according to the Diagnostic and Statistical Manual of Mental Disorders IV Text Revision (DSM-IV-TR), and were either initiating antidepressant monotherapy or were undergoing their first switch of antidepressant. Patients were excluded if they had schizophrenia or other psychotic disorders; bipolar disorder; substance dependence; dementia or other neurodegenerative disease significantly affecting cognitive functioning; or a mood disorder due to a general medical condition or substances. Patients receiving a combination of different antidepressant treatments at the time of the initial consultation were also excluded.

The physician’s choice of the drug used to treat each patient was based on clinical judgement alone and was not influenced by the decision to enter the study. All patients enrolled in the study provided written informed consent. The necessary ethical approvals were obtained for each study site before study initiation. The study was registered at ClinicalTrials.​gov: NCT01427439 on 26 August 2011.

Patient characteristics recorded at baseline included demographic information; the characteristics of the current episode of depression and – for patients switching antidepressants – the main reason for switching, according to the physician; history of depression; and the presence of mental health disorders other than depression, and of any functional syndromes.

Clinical profile at baseline was assessed by all patients using the 9-item Patient Health Questionnaire (PHQ-9) [17] and by all participating investigators using the Clinical Global Impressions–Severity of illness scale (CGI-S) [18]. The Montgomery–Åsberg Depression Rating Scale (MADRS) [19] was also administered when recruitment was conducted in the specialized sector and patients were evaluated by psychiatrists. The PHQ-9 scores were used to categorize the severity of symptoms as “minimal” (score 0–4), “mild” (score 5–9), “moderate” (score 10–14), “moderately severe” (score 15–19) or “severe” (score 20–27) for analysis. The CGI-S responses were analysed by the category assigned by the physician on a 7-point scale ranging from 1 = not ill, to 7 = extremely ill. The MADRS responses were analysed by total score.

Daily functioning and quality of life were assessed using self-administered instruments. Functioning was evaluated using the Sheehan Disability Scale (SDS) [20] including three dimensions (work, social and family functioning). The “activity impairment” dimension of the Work Productivity and Activity Impairment questionnaire (WPAI) was also applied in order to evaluate functioning [21]. The other dimensions of the WPAI are not presented as they aim to measure productivity and are relevant only to the subpopulation in employment. The Arizona Sexual Experience Scale (ASEX; five items, each scored from 1 to 6) [22] was used to assess sexual functioning. Patients were categorized as having sexual dysfunction if their total ASEX score was ≥19 or their score was ≥5 for one item or ≥ 4 for at least three items. Patient-reported outcomes were also used to assess health-related quality of life using the physical and mental health dimensions of the Medical Outcomes Study Short-Form (12-item) Health Survey (SF-12) [23] and the EuroQol 5-Dimensions questionnaire (EQ-5D; in UK patients only) [24]. Utility scores were derived from the EQ-5D scales by applying UK tariffs [25].

Resource use during the 12 weeks before baseline was assessed by physician reporting of physician visits, hospitalizations and periods of sick leave. Resource-use data were reported for patients who had experienced a depressive episode of at least 8 weeks’ duration as: per cent of patients reporting at least one visit (physician or other healthcare professional) or 1 day of hospitalization or sick leave over the recall period; number of visits (physician or other healthcare professional) or days of hospitalization or sick leave for patients with at least one visit or day of the corresponding resource item; and overall number of visits (physician or other healthcare professional) or days of hospitalization or sick leave. The restriction of this analysis to patients with a minimum depressive-episode duration of 8 weeks was applied to allow an appropriate comparison between the two treatment groups, given the between-group difference in duration of the current depressive episode.

Patients included in the analysable dataset were those who met the study selection criteria and completed a baseline assessment form as well as at least one post-baseline assessment or outcome questionnaire. Summary statistics (mean, standard deviation, median, minimum and maximum) were calculated for continuous variables, and numbers and percentages were calculated for categorical variables. Analyses were performed for both groups of patients: those switching and those initiating treatment. Differences between these groups were investigated using either Student’s t test for quantitative variables or chi-squared test (or Fisher’s exact test where appropriate) for qualitative variables. All analyses were univariate as the aim was to describe the characteristics of the study populations rather than to explain or identify any associations. Statistical significance was assumed at p < 0.05. Statistical analyses were performed using SAS® statistical software, Version 9.2 (SAS Institute, Cary, NC, USA).

A total of 1895 patients were screened, 1402 of whom were enrolled in the study. The first patient was screened on 25 February 2011 and the last patient completed the study on 19 February 2015. The most frequent reason for non-enrolment of screened patients (n = 493) was patient’s decision (n = 203; 41.2%). Of those who were enrolled, 1159 (82.7%) provided analysable data. Patients who were enrolled but not included in the analysable population (n = 243) were excluded because they violated at least one of the inclusion and/or exclusion criteria at baseline (n = 167) or they had not completed a post-baseline case report form or at least one post-baseline questionnaire in the predefined timeframe (n = 101, including 76 who met the inclusion criteria at baseline). The majority of patients were enrolled and followed up by general practitioners (n = 969; 83.6%). Overall the analysable population had a mean age of approximately 44 years, and the majority (n = 848; 73.2%) were female (Table 1). Included patients were from the UK (n = 341; 29.4%), France (n = 339; 29.2%), Spain (n = 270; 23.3%), Germany (n = 164; 14.2%) and Sweden (n = 45; 3.9%).

At baseline 910 (78.7%) patients were initiating antidepressant treatment and 247 (21.3%) were switching antidepressant for the first time; the treatment status of two patients was unknown. The main causes of switching (recorded for 216 patients) were lack of efficacy (n = 167; 77.3%), adverse events (n = 20; 9.3%), patient’s decision (n = 15; 6.9%) and lack of compliance (n = 7; 3.2%). Compared with patients who were initiating antidepressants, approximately twice as many patients in the analysable population who were switching antidepressants were being treated by psychiatrists at the time of enrolment (n = 124 vs 66, respectively; P < 0.001; Table 2).

There were statistically significant demographic differences between the two patient groups: compared with patients who were initiating treatment, those who were switching were significantly older (P = 0.012) and more likely to be widowed or divorced/separated (Table 1). Patients who were initiating antidepressant treatment were more likely to be educated to university degree level and were more likely to be in paid employment or self-employed compared with patients who were switching antidepressant treatment (Table 1).

With regard to their current episode of depression, the duration of the episode – classified in time categories – differed significantly (P < 0.001) between the two patient groups. Greater proportions of patients who were switching antidepressants than those initiating treatment had been experiencing the current episode of depression for more than 8 weeks (Table 2). In addition, a significantly higher proportion of patients who were switching treatment were experiencing symptoms of anxiety (P < 0.001); among patients with anxiety, those who were switching were significantly more likely to be treated with anxiolytics (P = 0.004).

Severity of illness, as rated by the patient using the PHQ-9, indicated little clinical difference between the treatment groups. Although the between-group difference in mean scores reached statistical significance (P = 0.014), the mean score was only marginally higher for patients switching antidepressants versus those initiating treatment (Table 2). Conversely, the difference in distribution by category of severity did not reach statistical significance (P = 0.084), despite a greater proportion of patients who switched treatment ranking their illness as severe (PHQ-9 score 20–27) compared with those initiating antidepressant treatment (Fig. 1).

The severity of the current depressive episode, as measured by the mean physician-assigned CGI-S score, was significantly higher in patients who were switching antidepressants compared with those initiating the treatment (P < 0.001) (Table 2). The profile of disease severity differed significantly in the two populations (P < 0.001), with a greater proportion of patients in the switching group rated as markedly, severely or extremely ill (48.4% versus 36.2% in the initiating group) (Fig. 2).

Compared with patients initiating antidepressant monotherapy, significantly greater proportions of patients switching antidepressants had experienced a previous episode of depression at any time (P < 0.001; Table 2).

No significant differences were seen with regard to the presence of comorbid mental disorders, but differences were noted for comorbid physical disorders: significantly higher proportions of patients who were switching had a diagnosis of a chronic pain (P < 0.001) or fibromyalgia (P < 0.001) compared with those initiating treatment (Table 2).

With regard to functioning as assessed by the SDS, the mean total SDS score was significantly higher – indicating greater functional impairment – in patients who were switching compared with those initiating antidepressants (P = 0.004) (Table 2). Furthermore, the individual SDS domain scores were also significantly higher in patients who were switching (each P < 0.05) (Table 2). The WPAI dimension of overall activity impairment showed a significant between-group difference, with a significantly higher mean score, i.e. worse impairment, in the group switching than in the group initiating antidepressants (P = 0.002; Table 2). The mean total ASEX sexual-function score was statistically significantly higher in the group switching than in the group initiating antidepressants (P = 0.031), although the numerical difference was small (Table 2).

The SF-12 mean physical component summary (PCS) score was significantly lower (indicating worse physical functioning) in the switching compared with the initiating group (P < 0.001); however the mean mental component summary (MCS) score did not differ between groups (Table 2).

Resource use within the 12 weeks before the baseline visit by patients with a depressive episode of greater than 8 weeks’ duration is summarized in Table 3. Compared with patients who initiated antidepressant medication, a statistically significantly greater proportion of patients who were switching antidepressants made at least one visit to any physician in the previous 12 weeks (88.0% vs 96.2%; P = 0.003). Patients who were switching antidepressants had more physician visits than patients initiating treatment in the same period (Table 3).

In terms of the type of physician visited, patients who were switching were also significantly more likely than those initiating treatment to have visited a psychiatrist (P = 0.001) or psychotherapist/counselling (P = 0.047) in the previous 12 weeks. A significantly greater proportion of patients switching antidepressants than those initiating treatment had taken at least one period of sick leave in the previous 12 weeks (P = 0.033).

The strengths of this study include the large sample size and its international scope. In addition, this is one of the first studies in which the collected baseline data comprise such a wealth of naturalistic information, including detailed individual medical histories and both physician- and patient-reported information on the current depressive episode, as well as functional, quality-of-life and resource-use data. The study population had the expected demographic characteristics of patients with depression: as previously reported in other observational studies, the majority of patients with depression were women, and the mean age of the population was approximately 45 years [11, 51, 52]. Although the categorization of depression severity using patient-reported outcome in the majority of patients in the present study (rather than by scales more frequently used in clinical trials) may be perceived as a limitation of the study, this approach is increasingly recognised as appropriate and may be considered a study strength. Greater focus on patient-reported outcomes reflects the general move towards increased patient involvement in treatment decisions, and awareness of the limitations of the more ‘traditional’ clinical symptom-based measures in assessing recovery from mental illness in a way that is meaningful to patients [53‐56]. In PERFORM a validated questionnaire, based on DSM-IV diagnostic criteria (PHQ-9) [17] was used to evaluate patient-reported depression, and the severity levels revealed by physician assessment using the CGI-S support the findings of the patient-reported assessment that patients switching antidepressants had more severe depression.

We acknowledge a number of study limitations. Patients were selected on the basis of having symptoms of depression, resulting in an over-representation in the switching population of patients switching treatment due to lack of efficacy (i.e. inadequate response to the previous antidepressant) compared with other reasons. The difference in depression severity between the two study populations may therefore be greater than would be seen in clinical practice; however, as drug-related adverse effects (the next most common reason for switching) may affect functioning and quality of life to a similar degree as a lack of efficacy, the potential bias towards selecting patients with more severe depression in the switching group may not have an effect on the findings. The study comprised outpatients only, potentially excluding some of the most severely affected patients. These criteria may have led to the study population of patients switching treatment having less severe depression than would be encountered in patients switching treatment in clinical practice, however, in Europe the vast majority of patients with MDD are treated in the outpatient setting [1, 57]. The study also excluded patients treated with combination or augmentation therapies. However, guidelines currently applied in Europe recommend initiating treatment with antidepressant monotherapy, with a switch to a different monotherapy in the event that an inadequate response is achieved with the initial therapy [8]. It would therefore be anticipated that in clinical practice few patients would receive a combination of antidepressants or antidepressant therapy augmented with another agent as either first- or second-line therapy. Because of these criteria, the outcomes may be more applicable to patients in the early rather than later period of their depressive episode. A degree of recall bias may also have occurred, which would have resulted in more extensive retrospective reporting of medical issues and resource use by the patients switching treatment. However, such patients would genuinely have experienced more clinical encounters as they had been treated for depression in the preceding weeks; as the recall period was restricted to 12 weeks, the potential for recall bias is low. Analyses of the MADRS and EQ-5D data were based on small sample numbers as these tools were only used in specific settings. This limits the statistical power of these analyses to detect between-group differences. However, the reasonably good response rate for other assessments of depression severity and quality of life – despite the fact that completion of the self-administered instruments was on an entirely voluntary basis – permitted reliable comparisons for these characteristics.