The burden of disease in patients eligible for mentalization-based treatment (MBT): quality of life and costs

In total, 403 patients were included between March 2009 and July 2014. Patients were included in these studies when they had a BPD diagnosis as measured by the Structured Clinical Interview for DSM-IV Axis II Personality Disorders (SCID-II) [27]. In the MBT-TAU study, patients should also have a total score on the Borderline Personality Disorder Severity Index (BPDSI) [28] of at least 20, reflecting severe BPD. Exclusion criteria were: (a) the presence of an Axis-I disorder (as determined with the Structured Clinical Interview for DSM-IV Axis I Personality Disorders (SCID-I) [29] that required specialist treatment, (b) a diagnosis of autism spectrum disorders or organic brain disorder that interferes significantly with the ability to mentalize; (c) IQ below 80 as measured by the Wechsler Adult Intelligent Scale–III [30], (d) inadequate mastery of the Dutch language, and (e) a diagnosis of antisocial personality disorder in combination with a history of severe physical violence [25, 26].

The SCID-I and SCID-II interviews were administered as part of the formal admission procedure (De Viersprong) or to assess whether patients were eligible for the trial (Arkin and Lentis). If patients fulfilled the inclusion criteria for one of the trials, patients signed informed consent and were asked to complete a baseline assessment, including the EQ-5D-3L and TiC-P measurements. In the MBT-DOS trial, patients completed the baseline assessment before randomization. In the MBT-TAU study, patients completed the baseline assessments partly before and partly after randomization.

The Dutch version of the Brief Symptom Inventory (BSI) [34, 35] was used to assess general psychopathological symptoms. The BSI is the short version of the Symptom Checklist-90. It includes 53 items covering nine symptom dimensions (somatization, obsession-compulsion, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation and psychoticism) and yields three global indices of distress: Positive Symptom Distress Index, Positive Symptom Total, and Global Severity Index (GSI). Possible GSI scores range from 0 to 4, with higher scores indicating a higher level of psychological and emotional distress. Respondents have to rate each item (e.g. “your feelings are easily hurt”) on a five-point scale ranging from 0 (not at all) to 4 (extremely), representing the intensity of distress relating to each item during the past 7 days. The reliability of the Dutch version of the BSI is good (Cronbach’s α ranging from .71 to .88, test-retest reliability ranging from r = .71 to .89). These values are comparable to the original BSI version of Derogatis [34].

QoL was measured using the EuroQol EQ-5D-3L [36]. This self-report questionnaire provides a simple method to measure health problems in 5 dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. Each dimension is divided into three response-levels: no problem, moderate problems, and extreme problems. Combining these scores gives 243 possible health states, where each of these is weighted to obtain a single index score between -.33 (worst imaginable health state) and 1 (best imaginable health state) [37]. A score of 1 represents the value of perfect health, whereas a score of 0 represents the value of death. Sometimes patients experience extreme problems on all five dimensions, which can turn their QoL index score below zero. However, normally this state is temporary [38]. To calculate the mean QoL index score, we used norm scores of a Dutch validation study [37]. The QoL index score is thus estimated based on the preferences of the Dutch general public and represents a value the public would assign to the measured health state [39]. The reliability of the EQ-5D-3L has been investigated and found to be acceptable [40] and it has shown to be sensitive to change in patients with personality disorders [22, 41].

The direct costs (e.g. increased health care utilization) and indirect costs (e.g. productivity losses due to disability and to higher absence from work) in the year prior to treatment were assessed using the Trimbos and Institute for Medical Technology Assessment Questionnaire for Costs Associated with Psychiatric Illness (TiC-P) [42, 43]. The first part of the TiC-P consists of questions on (1) the number of visits to, e.g. a general practitioner, psychiatrist (outside the treatment offered in the trials), medical specialist, physiotherapist or alternative health practitioner; (2) the day care/hospital lengths of stay (outside treatment offered in the trials); and (3) the use of medication in the 4 weeks prior to the administration of the questionnaire in MBT-DOS and 6 months in MBT-TAU [42]. These care and cure volumes were multiplied with unit prices of the corresponding health care services according to the Dutch manual for costing studies in health care [44]. The unit prices for 2010 will be adjusted to 2014 prices using the Consumer Price Index [45]. Both recall times (4 weeks in MBT-DOS and 6 months in MBT-TAU) were standardized to 1 year.

The second part of the TiC-P asks the patient to report absence from work, reduced efficiency at work and difficulties with nonpaying jobs. The days of short-term absence from work and actual hours missed at work due to health-related problems were multiplied with the average productivity costs per employee per hour, taking into account the number of days and hours of paid employment of the patient per week. Because the recall period of this part of the TiC-P was 2 weeks, we multiplied the costs by 26 to estimate the annual costs. To value long-term absence from work, we applied the friction cost method [22, 46, 47]. This method takes the employer's perspective, and counts as lost only those hours not worked until another employee takes over the patient's work. This period, the so-called friction period, is estimated to be 116 days. Hence, the maximum indirect costs were limited to productivity losses during 116 days. The friction cost method is a more conservative estimate than the so-called “human capital method”, which relates productivity costs one-to-one to the labour costs of the patient. The choice between friction costs and human capital is still a subject of debate among economists [22, 46, 47]. Furthermore, difficulties with non-paying jobs were estimated as hours a patient was supported in for example housekeeping activities, and was multiplied with the corresponding unit prices [42, 43].

We estimated the burden of our patient population in terms of their QoL and direct and indirect costs, which represent the societal costs of these patients. In addition, we described their socio-demographic and diagnostic data and GSI scores. We reported the mean and standard deviation of the QoL index score, the direct and indirect costs, and the GSI score.

Table 1 presents the social demographic and diagnostic data at baseline. As can be expected, the majority of the sample was female (82 %). The majority (81 %) of the patients had at least one co-occurring Axis-I disorder and 41 % had at least one co-occurring Axis-II disorder. The mean GSI score at baseline was 1.92 (SD = .79); representing a high level of psychological and emotional distress.

The mean QoL index score in our patient population was .48 (SD = .29), representing a severe burden of disease (Table 1). Table 2 describes the direct and indirect costs in the year prior to treatment. The mean total direct costs per patient were €13,405 per year. Those costs were mainly composed of costs due to inpatient health care (39.1 %), outpatient mental health care (18.6 %) and day hospital care (14 %). The total mean indirect costs per patient were €3474 per year. Here, absence from work and reduced efficiency at work were estimated for the 90 patients (28.1 % of 320 valid N; 83 missing values) that had a paying job. These patients reported the total hours lost to be 101.34 (Table 2). The mean total costs of BPD patients were €16,879 per year.

One of the strengths of this study is the use of a generic QoL measure, which made it possible to compare the QoL index score of the BPD patients in our study to that of patients with other mental and physical illnesses. Although it is suggested that the EQ-5D-3 L may not be sensitive enough to reflect the impact of severe mental disorders, such as chronic schizophrenia and PDs [54], the current study showed that these concerns are not justified for severe BPD patients in our study population. We found a severely disturbed QoL in these patients, which indicated that an important part of the problems in this particular patient group was well captured with the EQ-5D-3 L. In addition, other studies on QoL in BPD patients found similar QoL index scores, indicating the robustness of the present findings [12, 55]. A second strength was the use of an extensive cost questionnaire, by which we could calculate both direct and indirect costs. Bateman and Fonagy only estimated direct costs, and did not take into account the indirect costs due to productivity losses [23]. The current study now added those costs to the total costs of patients eligible for MBT, and made costs comparable to other studies on the economic burden of PD in general and BPD specifically.

We only included BPD patients eligible for MBT, which may limit the generalizability of our study results to the general group of BPD patients. However, our results showed that societal costs and QoL were comparable to other studies on QoL and costs in BPD patients. A second limitation is the considerable number of missing data, especially in the cost data, as can be seen in Table 2. We mostly missed data from the RCT that aimed to compare MBT-DH with S-TAU. This could be explained by the procedure of measurements. Only the SCID-II and the BPDSI were calculated for all patients before randomization as part as the admission procedure for this study. The other baseline measures were administered after randomization, including the EQ-5D-3L, TiC-P and BSI. As a lot of the S-TAU patients were unhappy with their treatment allocation, many of these patients refused to complete the test battery after randomization. Though other studies on costs in BPD patients do not mention missing data [21‐23], we know out of personal communication that the study by Soeteman also had a considerable number of missing data. Despite the missing data in our trial, the QoL and EQ-5D score were still within the same range as these earlier studies. Missing data influences, however, the costs estimates as presented in Table 2. On the one hand, the patients whose costs were unknown could have had a high amount of health- related costs, such that the overall estimated costs represent an underestimation. On the other hand, it is possible that these patients had no costs at all. In that case, the estimated costs in this study could be overestimated, because we only represented the costs of those patients that actually filled in the cost questionnaire. A third limitation relates to the measurement of costs by self-report. The costs were not based on more objective sources such as data from health insurance companies, which may have led to an underestimation of costs. However, self-report was commonly used to calculate costs on BPD patients in earlier studies, and to our knowledge, there are no studies using data from health insurance companies yet. In addition, data from health insurance companies is limited to costs covered by the health insurance and has no information on, for example, out of pocket costs. A fifth limitation is that we calculated the costs prior to start of treatment. These costs may be very high, because patients had severe problems, for which they eventually received treatment. On the other hand, it is also possible that costs prior to treatment are lower than usual, because patients know they will receive treatment and therefore don’t make use of other resources.

Our study showed a high burden of disease in BPD patients eligible for MBT. This high burden can help prioritizing health interventions for severely ill BPD patients in general, and MBT specifically, as people tend to be willing to pay more to treat patients with a higher burden of disease. Given the low QoL and high societal burden we found in this study, an expensive treatment for severely ill BPD patients, such as MBT, can more easily be legitimized. However, as the burden of disease in severe BPD patients is comparable to the burden of other BPD patients and patients with other PDs, prioritization of resources for severe BPD patients based on (cost)-effectiveness research becomes even more important. Though an expensive treatment is more easily legitimized, our results provide no license to unlimitedly use resources to reimburse MBT or other highly specialized treatments for severe BPD patients, as these findings do not provide any information on the (cost-)effectiveness of available treatment programs. The effectiveness of treatment programs should be evident, not only by studies on the effectiveness of the treatment itself, but also by comparing the effectiveness of MBT to TAU and to other highly specialized treatments for BPD patients. Although there is some promising evidence supporting the effectiveness of MBT-DH and MBT-IOP [3, 56‐58], given the small number of studies, more research is urgently needed, particularly in light of the limitations of existing trials. These include possible researcher allegiance, questions about generalizability and the lack of a credible TAU. Moreover, more research into the effectiveness of other specialized treatments for BPD, such as SFT and DBT, is also warranted. Various studies show that the effectiveness of these highly specialized treatments seems comparable to well-specified treatments for BPD that are delivered in a consistent, coherent and continuous way [59‐61]. As a result, in practice, integrative treatments are already being developed [62], combining the best of specialist techniques and common factors. Yet, research on the effectiveness of these integrated treatments is needed.

Furthermore, when for example MBT turns out to be more effective, but also more costly than the intervention to which it would be compared, a cost-effectiveness analysis of MBT is needed to estimate the cost-effectiveness ratio of MBT compared to alternative treatment options. Currently, we are comparing day hospital MBT to specialist TAU [25] on its effectiveness and costs. At the same time, two intensities of MBT, day-hospital MBT [56] and intensive outpatient MBT [57], are compared on effectiveness and costs [26]. Cost-effectiveness analyses can support the reimbursement decision in which treatment should be paid for. However, a cost-effectiveness analysis is not the only tool to decide on this. The burden of disease is also an important criteria to decide which treatment should be paid for, and can yield a strong argument in favor of reimbursing treatments for severely ill BPD patients.