Comparative economic evaluation of quetiapine plus lamotrigine combination vs quetiapine monotherapy (and folic acid vs placebo) in patients with bipolar depression (CEQUEL)

Abstract Objectives Although not licensed for acute bipolar depression, lamotrigine has evidence for efficacy in trials and its use is recommended in guidelines. So far there had been no prospective health economic evaluation of its use. Methods Cost‐utility analysis of the CEQUEL trial comparing quetiapine plus lamotrigine vs quetiapine monotherapy (and folic acid vs placebo in an add‐on factorial design) for patients with bipolar depression (n = 201) from the health and social care perspective. Differences in costs together with quality‐adjusted life years (QALYs) between the groups were assessed over 52 weeks using a regression‐based approach. Results Health‐related quality of life improved substantially for all randomization groups during follow‐up with no significant difference in QALYs between any of the comparisons (mean adjusted QALY difference: lamotrigine vs placebo −0.001 (95% CI: −0.05 to 0.05), folic acid vs placebo 0.002 (95% CI: −0.05 to 0.05)). While medication costs in the lamotrigine group were higher than in the placebo group (£647, P < 0.001), mental health community/outpatient costs were significantly lower (−£670, P < 0.001). Mean total costs were similar in the groups (−£180, P = 0.913). Conclusions Lamotrigine improved clinical ratings in bipolar depression compared with placebo. This differential effect was not detected using the EQ‐5D‐3L. The additional cost of lamotrigine was balanced by significant savings in some other medical costs which made its use cost neutral to the health service. Compared to placebo, folic acid produced neither clinical nor significant health economic benefits. The study supports the use of lamotrigine in combination with other drugs to treat bipolar depression.

In fact, the adoption of lamotrigine has lagged behind that of other new drugs used in bipolar disorders in the UK. 13 This probably reflects the absence of a marketing authorization for its use in bipolar depression. Therefore, evidence for efficacy and cost-effectiveness from independent clinical trials is of critical importance in improving practice. The analysis of further data from CEQUEL on quality of life, costs and cost-effectiveness over 52 weeks of treatment of patients with bipolar disorders here contributes a complementary perspective for considering the regular use of lamotrigine.

| Study design and study population
CEQUEL was a multi-centre (27 sites), double-blind, randomized, placebo-controlled, parallel group, 2 × 2 factorial clinical trial con- and without contraindications to do so were assigned to folic acid (500 µg/d) or placebo folic acid. Out of the 202 study participants, 94 participants were thus separately randomized to folic acid and 92 participants to placebo folic acid. Sixteen participants were not randomized to the folic acid/placebo comparison. One participant died during the 52 weeks follow-up (suicide; group allocation: placebo lamotrigine/active folic acid) and was excluded from the health economic analysis. Table 1

| Outcomes
The primary economic analysis was an incremental cost-utility analysis.
Outcomes were expressed as QALYs and calculated as the area under the curve based on the EQ-5D-3L data. EQ-5D-3L is a standardized, non-disease-specific instrument designed for valuing general healthrelated quality of life. 24 It is a generic, self-reported measure of healthrelated quality of life alongside five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) which allows the preference-based, comparative evaluation of distinctively different treatment options and disease areas and is therefore the preferred outcome measure to be used in economic evaluations by NICE. 25 Although EQ-5D was found sensitive to changes in health-related quality of life of bipolar patients during their depressive episodes, its sensitivity to change in episodes of mania or mixed episodes however remains unclear. 26 EQ-5D utilities were based on the UK tariff values. 27

| Costs
UK national-level unit costs were applied for each resource use item to calculate total costs of resources used (

| Analyses
In line with NICE guidelines, cost-effectiveness was primarily assessed from the health and social care perspective. 25 To account  for the high indirect costs associated with bipolar disorders, 28 indirect costs (lost productivity costs) were also measured (Supporting Information Table S2).
To accommodate for the factorial design of the trial and to be in line with the main clinical analysis, a regression-based approach using ordinary least squares (OLS) was adopted to analyse mean outcomes and mean costs by randomization group. This technique not only allows adjustment for the effect of the second intervention (ie, folic acid in the analysis on lamotrigine and lamotrigine in the analysis on folic acid), but also to control for covariates and to deal with different sample sizes between groups. 29 As some patients chose All analyses were done on an intention-to-treat basis. A p value less than 5% was considered as statistically significant. Microsoft ® Excel 2013 was used for costing and Stata ® 13.1 for the statistical analyses.

| Outcomes
EQ-5D utility analyses were based on two main approaches: available cases analysis and full dataset analysis following multiple imputa- incomplete cases. 31 A regression-based approach adjusting for randomization group, baseline EQ-5D utility, age and sex was applied to estimate mean EQ-5D utility values and QALYs (Table 4). 32 In the main analysis, transitions in health-related quality of life between 0 and 12 weeks as well as 12 and 52 weeks were assumed to be linear.
Further sensitivity analyses were carried out assuming the health

| Costs
For inpatient resource use and medication, data for the full sam- Finally, the potential interaction effects between randomization groups were investigated for all participants randomized to the lamotrigine and folic acid comparison (Supporting Information Table S3) (n = 185).

| Cost-effectiveness
The main analyses of QALYs (Table 4) and total health and social care costs (

| Outcomes
Both the available EQ-5D utility values and the full imputed EQ-5D dataset for baseline, 12 weeks and 52 weeks follow-ups showed substantial improvement for all randomization groups during the 52 weeks follow-up period (Table 3).
Regression adjusted mean EQ-5D utility and QALY differences between randomization groups are given in Table 4. No betweengroup differences were seen for any of the group comparisons.
Participants taking lamotrigine had slightly lower QALYs (−0.001; P = 0.972) than participants taking placebo, whereas participants randomized to folic acid had slightly higher QALYs (0.002, P = 0.938) than those allocated to placebo but these differences were neither statistically nor clinically significant. Relevant sensitivity analyses of the QALY calculation methods confirmed the main results and are therefore not presented separately.  and 48% of the total health and social care costs in the lamotrigine and placebo arms, respectively. Overall, inpatient costs were reported for 24 out of the 201 analysed participants (lamotrigine placebo: 14 participants; lamotrigine active: 10 participants) and were somewhat but not significantly higher in the placebo group (Table 5).

| Costs
Folic acid was associated with lower mean total costs (n = 91, £5,438, SE: £1,268) than placebo (n = 94, £6,952, SE: £1,248) when controlled for lamotrigine, but the difference was not statistically significant (−£1,514, P = 0.397, n = 185; Table 5). The effect was dominated by the inpatient costs associated with one individual in the placebo group incurring total hospital costs of £97,670 during the 52 weeks follow-up period. Numbers incurring inpatient costs were comparable (folic acid placebo: 11 participants; folic acid active: 13 participants). There were no statistically significant differences in any of the individual cost items.
Lost productivity due to work absence was analysed based on the available data. Of 119 participants, 41 were in employment at baseline and an additional six participants started working during the trial (n = 47). Of these 47 participants, 35 reported absenteeism due to (any) sickness during the follow-up period for a total of 1,266 work days. Overall, the mean cost of lost productivity per patient in the placebo group significantly exceeded the lamotrigine group by £2,755 (P = 0.037), whereas no statistically significant difference was identified in the folic acid comparison.
Three outliers in terms of inpatient and overall health and social care costs were identified (two participants: active lamotrigine, active folic acid; one participant: placebo lamotrigine, placebo folic acid). Sensitivity analyses excluding these three participants confirmed the conclusions from the main analysis.

| D ISCUSS I ON
In this economic evaluation of the effects of adding lamotrigine and folic acid to quetiapine therapy to treat bipolar depression, quality of life improved substantially for all groups during follow-up with no significant differences in health outcomes between any of the group comparison. The same applies to costs, indicating no significant differences between groups in terms of total health and social care costs. Overall, this analysis suggests that the addition of lamotrigine to quetiapine is associated with good health outcomes and is cost neutral. Compared to placebo, folic acid produced neither clinical nor significant health economic benefits.

| Quality of life
The effect of treatment of bipolar depression has so far not been systematically measured from a health economic perspective. The change from around 0.5-0.7 for quality of life is both statistically and clinically significant. Regular mood monitoring may have contributed to this overall positive outcome as suggested by the earlier findings of Bopp and colleagues on steady improvement in patients using TrueColours. 34 On the other hand, the EQ-5D measure did not discriminate lamotrigine from placebo in the way that symptom ratings did. This is despite the fact that the EQ-5D correlates quite well with depressive ratings on the QIDS scale in bipolar patients. 25 Most likely, the EQ-5D may also be confounded by an uncertain relationship to manic symptoms, because they may be viewed in a paradoxically positive way, whenever they occur. This makes it a potentially less reliable outcome measure for bipolar disorder than for major depression. This hypothesis on the responsiveness and reliability of the EQ-5D in bipolar disorder needs further investigation. in another systematic review. 34 They are, however, not comparable to this cost-utility analysis as none of them investigated the effect of lamotrigine and quetiapine combined, while healthcare system differences in the included studies may also contribute to these cost variations. 34 Based on the results of this study, the lower mean productivity loss for the lamotrigine group suggests that quetiapine and lamotrigine combination therapy could result in further patient benefits and cost savings for the society. Given that bipolar disorders most commonly start in the early adulthood and the highest disease burden occurs in the working-age population, 1 this further emphasizes that the societal cost of lost productivity should be assessed in future economic evaluations.

| LI M ITATI O N S
Firstly, due to difficulties in collecting information on informal care utilization using the TrueColours instrument, these indirect costs could not be incorporated in the sensitivity analysis from the broader societal perspective. Secondly, with full information over the 52 weeks follow-up period for medication and inpatient use only, data on other health and social care use for 82 participants had to be fully and for 38 patients partially imputed. A sensitivity analysis based on available cases data (n = 119), however, supports the conclusions from the main analysis (Supporting Information Table S2). Furthermore, the reasons for drop out in the CEQUEL trial were the same in the different arms as were the numbers. 9 Thirdly, the aim of the clinical study was to balance the group allocation for bipolar disorder type, age and sex (Table 1). In terms of health-related quality of life, however, it seems noteworthy that some imbalances existed. For the group not allocated to the folic acid comparison, mean EQ-5D utilities at baseline (0.427, SD = 0.315) were (statistically non-significantly) lower than in all other groups. This suggests that opting out from the folic acid component of the study correlated with a lower quality of life. All outcome analyses were adjusted for EQ-5D baselines utilities to adjust for any relevant differences. Fourthly, although total hospital costs were found to be non-significantly lower in the lamotrigine group and in the folic acid group than in the placebo groups, these results are not robust due to the insufficient sample and event sizes for such inferences (inpatient resource use was only reported for 24 participants in total). Finally, the trial was not powered to detect interaction effects. A sensitivity analysis (Supporting Information Table S3) including an interaction dummy for both lamotrigine and folic acid combination therapy, however, suggests overall higher mean health and social care costs for participants taking quetiapine, lamotrigine and folic acid combined and thus supports the original clinical findings on the negative impact of concomitant lamotrigine and folic acid therapies. Finally, variables such as educational level, income, occupation, marital status, number of previous episode will also affect outcomes, but there is no reason to expect an impact on lamotrigine response per se.

| CON CLUS IONS
This health economic analysis complements and amplifies the clinical findings 9 and has important clinical and policy implications in terms of supporting calls for a wider use of lamotrigine to treat bipolar depression in published consensus guidelines.

ACK N OWLED G EM ENTS
We thank all those who contributed to the CEQUEL study economic