9.1. METHODS OF ECONOMIC EVALUATION

Methods of economic evaluation command a fairly high level of consensus and are reported in Drummond and colleagues (1997). However, where economic evaluations have been undertaken for anxiety disorders generally (see Issakidis et al., 2004 for example), costing data for direct and indirect costs attributable to OCD are scarce¹⁶. Part of the issue is that OCD is a chronic, relapsing and remitting condition, where the result of treatment is modest. According to Bobes and colleagues (2001), OCD ranks tenth in the World Bank and World Health Organisation's leading causes of disability. The overall efficiency at which treatments are delivered is an important consideration when resources are limited. Beyond the healthcare perspective, the economic burden related to OCD comprises not only direct medical costs but also costs due to premature death, unemployment and reduced productivity when the condition is left untreated. Efficient service utilisation based upon rigorous health economic evaluations of OCD would reduce the social and economic burden of this condition, which would ensure optimal healthcare is delivered within the constraints of the national budget. Since OCD often starts in and continues beyond childhood and adolescence and in time increases in severity if left untreated, its economic burden accrues in the total direct and indirect costs that accumulate with time. The direct costs and cost effectiveness over a time course of 1-year are the focus of this section.

9.2. USE OF HEALTH SERVICE RESOURCES

Using the human capital approach, DuPont and colleagues (1995) estimated the direct and indirect costs of OCD to be $8,400 million in 1990 USD prices, 6% of their estimated $147,800 million cost of all mental illness in the United States. The same study estimated the indirect cost component of OCD, due to lost productivity, which amounted to $6,200 million, or 74% of the total estimated cost cited above (Dupont et al., 1995).

9.3. PRIMARY CARE DRUG THERAPY VERSUS SECONDARY CARE CBT VERSUS COMBINED CBT PLUS SSRI THERAPY

There are a number of medications that are routinely administered in a primary care setting for the initial treatment of OCD. Traditionally, the most expensive of these has been the SSRIs. However, the differences in cost between alternative treatments need to be evaluated in light of their net effects beyond placebo. To accomplish this, one needs to know both the cost and net effectiveness of each treatment. For example, a 1-year course of the SSRI paroxetine (20 mg tablet in generic form) would cost an estimated £289, including three GP prescribing sessions¹⁷ and follow-up costs,¹⁸ which amounts to ±£10 in comparison with other generic drugs.

A meta-analysis was undertaken to obtain the weighted averages (according to number of patients in each study) of trial-based effectiveness data, based on 10 studies¹⁹ reported in this guideline. The results suggest that approximately 43% of OCD patients respond following a 12-month course of SSRI therapy, in comparison with 27% in the placebo group. This translates into a 16% treatment-related response rate. Four sensitivity analyses are included in this section to show how cost effectiveness may change in light of changes to effectiveness assumptions, such as those estimated by experts in the GDG. For clinician-guided CBT, the results from Greist and colleagues (2002) and Cordioli and colleagues (2003) suggest that 56% and 70% of OCD patients respond following a 12 month course of CBT, in comparison with 14% (systematic relaxation) and 4% (waitlist control) of controls. Therefore, the average treatment-related response rate for CBT was 53%, meaning that 53/100 patients respond, due to the treatment alone, following clinician-guided CBT²⁰. The treatment-related response rate of combined SSRI and clinician-guided CBT plus ERP is estimated to be slightly higher than these therapies alone, with 28 and 32% more respectively than SSRI and CBT as monotherapies. When the response rates of the individual treatments are averaged it is estimated that the overall response rates to combined treatment would be 63% within 3–12 months²¹. In comparison with SSRIs, in the short term, treatment costs are known to be higher with clinician-guided CBT because such therapy is labour intensive and requires specialist knowledge for optimal delivery. The GDG estimated that approximately 16 1-hour sessions of CBT would be a realistic frequency and duration of therapy. Delivered by a clinical psychologist, the sum of these sessions will cost £1,056²², or £767 more than the above SSRI monotherapy regimen.

Component costs and baseline effectiveness estimates are reported in Table 3. There are variations in both cost and effectiveness in comparing alternative monotherapies with their combination. To illustrate these differences, incremental cost-effectiveness ratios (ICERs) are compared in Tables 4 and 5, with GDG-suggested comparisons reported in parentheses.

Table 3

Cost and effectiveness estimates of alternative treatments for OCD.

Table 4

Incremental cost-effectiveness analysis: ICERs estimated for the nominal scenario (response rates = 16% for SSRI, 53% for CBT, and 63% for combined); assuming population of 100 patients treated.

Table 5

Sensitivity analysis 1: estimations of cost-effectiveness assuming estimates of the GDG (response rates = 50% for SSRI, 50% for CBT, and 60% for combined therapies); assuming population of 100 patients treated.

9.4. INTERPRETATION

Comparing the treatments for OCD presented in Table 3, the average cost per treatment-related response was £1,806 for SSRI therapy, £1,992 for CBT, and £2,135 for combined SSRI plus CBT. In all cases the average cost per treatment-related response falls within generally accepted limits of cost effectiveness. Even if the response rates were .10 higher or lower, they would still fall within these limits. However, these average costs are unsuitable for decision making and it is important to consider the incremental cost effectiveness between the various treatments.

In the nominal scenario, CBT is dominated by (that is, has an inferior cost-effectiveness ratio) a linear mixture of SSRIs and combination therapy. In other words, this is a case of ‘extended dominance’ where treating the entire presenting population with CBT alone is more expensive and less effective than treating part of the population with SSRIs and the remainder with combination therapy. This means that the only relevant comparison is the incremental cost effectiveness of SSRI therapy versus combined CBT + SSRI therapy, which costs £2,247 to achieve an additional treatment-related response. Dividing this ICER value by the NICE cost effectiveness lower-threshold of £20,000 per QALY yields 0.11 of a QALY. This value is the minimum QALY gain per response that would be required for combined therapy to become cost effective in comparison with SSRI therapy alone, at the lower threshold. At a cost-effectiveness threshold of £30,000 per QALY (NICE upper-threshold) an additional responder would have to be worth 0.07 of a QALY (that is, a 7% gain in quality of life).

In terms of EuroQOL, a standardised and self-reporting mechanism of rating health outcome that is weighted to the preferences of the UK population, a move from level 2 (moderately anxious or depressed) to level 1 (not anxious or depressed) for 6 months represents a gain of at least 0.035 of a QALY, or a 3.5% gain in quality of life. In the case of OCD, where the response to treatment translates into a gain of at least a 7% gain in quality of life, a cost-effectiveness ratio of £2,247 to achieve an additional treatment-related response is likely to represent good value for money. Under the present assumptions, combined therapy is likely to be a cost-effective option. If at least an additional 0.07 of a QALY is gained per additional response, combination therapy will be more cost effective than SSRI therapy alone.

There will be sub-groups for whom SSRIs are not an option because they are contraindicated, ineffective for whatever reason, or because individuals refuse to take them. In such instances, CBT alone may be cost effective and further analyses will need to consider the ‘do nothing’ option compared with CBT alone. The results reported in these tables did not include ‘do nothing’ options for each of the three treatments because the control groups were not comparable across studies. Further, the estimates of effectiveness for the CBT alone and combined SSRI and CBT treatments are based on very few studies and the control conditions are problematic. Consequently, the scenarios should be considered as illustrative only until more data from better controlled studies are available.

This model of 100 individuals did not consider the fact that patients often remain on drugs for much longer than a 12-month course of CBT, which will have an impact on the cost effectiveness. No value has been placed on patient choice: it has been assumed that all treatment options are equally acceptable to all patients. Another limitation is that response rates and response thresholds differed between the studies on which the weighted averages of effectiveness were based. Some utilised a 25% Y-BOCS threshold and others a 35% Y-BOCS score, for example. Others took the CGI as their scale of comparison. Future research should aim to utilise comparable scales of analysis and comparable samples of patients and presentations.

It is important to note that this model only considers up to a 12-month window of therapy. Quality of life gain due to response in the first and subsequent years must be weighed against the likelihood of relapse or a decline in response in the first and subsequent years. It may be that CBT, whether delivered as a monotherapy or as combined therapy, provides a more lasting response and/or less chance of relapse when medication is withdrawn. If there is a sufficient quality of life gain from response within 12 months to justify combined CBT plus SSRI therapy compared with SSRI therapy alone, any QALY gain beyond 12 months would only strengthen the cost effectiveness of this option.

9.4.1. Multi-way and extreme scenario sensitivity analyses based on GDG estimations and extreme effectiveness assumptions

Based on expert approximations of the GDG, the response rates for SSRI therapy, clinician-guided CBT and combined SSRI plus CBT therapy were estimated at 50%, 50%, and 60%, respectively. These estimates differ from approximations based on published data and are included in Sensitivity Analysis 1 (Table 5). If both SSRIs and CBT achieve a 50% treatment-related response rate, then simple cost-minimisation analysis will favour SSRIs as a first-line treatment because they are significantly cheaper. Thus, inclusion of CBT in this analysis is meaningless as it is dominated by SSRI therapy.

Using the GDG assumptions (reported in the parentheses of Table 3 that form the basis of Table 5), the incremental cost effectiveness of combined CBT plus SSRI therapy compared with SSRI monotherapy amounts to £2,890 per additional treatment-related response. Dividing this ICER value by the NICE cost effectiveness lower threshold of £20,000 per QALY yields 0.14 (14%) of a QALY, which is the minimum gain per response that would be required for combined therapy to become cost effective in comparison with SSRI monotherapy. At the upper-threshold of £30,000 per QALY, an additional responder would need to be worth 0.10 of a QALY, or a 10% gain in the quality of life, which is double the amount of gain required under the assumptions in Tables 3 and 4.

Tables 6–8 show extreme scenario analyses that demonstrate how sensitive cost effectiveness results are at present to changes in the effectiveness values.

Table 6

Sensitivity analysis 2: extreme scenario estimations of cost-effectiveness assuming equal treatment-related response rates of 25% between all treatments; assuming population of 100 patients treated.

Table 7

Sensitivity analysis 3: extreme scenario estimations of cost-effectiveness assuming linear 10% increase in response rate with SSRI, CBT and combined treatments (response rates = 30% for SSRI, 40% for CBT, and 50% for combined therapies); assuming population (more...)

Table 8

Sensitivity analysis 4: extreme scenario estimations of cost-effectiveness assuming linear 10% decline in response rate with SSRI, CBT and combined treatments (response rates = 50% for SSRI, 40% for CBT, and 30% for combined therapies); assuming population (more...)

If all treatment-related response estimates were equal between the comparators at 25% each, Table 6 shows that a cost-minimisation situation would arise where it would not be cost effective to offer the comparator options since they would provide no benefit and accrue only a cost.

Table 7 assumes that the treatment-related response rates for SSRIs, CBT and combined treatments are 30%, 40% and 50%, respectively. In this case, as in the case of Table 4, CBT is dominated by a linear mixture of SSRIs and combination therapy. The cost effectiveness of SSRI compared with combined treatment is therefore the only meaningful comparison. At the upper-threshold of £30,000 per QALY, an additional responder would need to be worth a 35% gain in the quality of life, which is seven times the amount of gain required under the assumptions in Tables 3 and 4.

If the treatment-related response rates for SSRIs, CBT and combined treatments are inversed from the scenario in Table 7, then, as depicted in Table 8, SSRIs would become dominated by a mixture of CBT and combined therapies. Combined treatments compared with SSRIs would then be the only meaningful comparison and a response would need to be worth a 35% gain in the quality of life in order for SSRIs to be cost effective in comparison with combined treatments.

In summary, the results depicted in this section are highly sensitive to changes in the effectiveness assumptions and future studies will need to establish confidence in these estimates of effectiveness before any estimates of cost effectiveness can be considered robust. In the meantime, it is anticipated that Tables 3–8 can serve as a template which can be continuously updated as new data become available on both cost and effectiveness.

9.5. NON-HEALTHCARE BURDEN

Whereas 0.5–2% may have OCD in the background population, only 25% of sufferers actually present to a GP and fewer still continue through the full course of treatment when such is made available. OCD also presents a considerable economic burden to the individual, family, health services, and society as a whole. The total cost accrued as a result of OCD-related illness is difficult to measure, because it extends beyond the primary, secondary, and tertiary care settings. Often sufferers accrue costs attributed to work-related absences, and beyond the immediate family. These latter costs may arise from the affected individuals as well as from family and friends who care for them.

A survey of an OCD consumer advocacy group estimated that, on average, a person with OCD loses fully 3 years of wages over their lifetime (Hollander & Wong, 1995). If an OCD sufferer incurs losses of £483.04 for every week they are absent (Income data services, 2004), this would amount to a total of £75,354 due to unemployment over this 3-year period, not including lost opportunities for career advancement and the cost to families and carers over their respective working lifetimes. The long-term, societal costs are beyond the NICE scope of the healthcare perspective. Nonetheless, they highlight the importance of delivering interventions at the earliest signs of illness and strengthen the conclusion that combination therapy may be a cost-effective option.

9.6. CONCLUSIONS AND FUTURE RECOMMENDATIONS

The analyses presented in this section assume a presentation of OCD that is, on average, of moderate intensity. More severe presentations will obviously require more intensive treatments over a longer duration of time. Further analyses are needed to compare the cost effectiveness of the alternative therapies over realistic time courses. Also, more research is required to compare the stepped-care approaches with the monotherapies, when treatments are administered in their combination over different time courses.

The analyses presented in Tables 3–5 suggest that combined CBT plus SSRI therapy is a cost-effective option for individuals suffering from OCD who might otherwise receive only SSRI therapy. If the quality of life gains are not achieved at an acceptable incremental cost, however, SSRI would be the cost-effective option. CBT alone is unlikely to be cost effective with the assumptions reported in either Table 3 or Table 4. Delivery formats for CBT which require less therapist input per patient treated would result in lower treatment costs, and, depending on response rates, may prove to be cost effective. For individuals who are unable or unlikely to remit following SSRI therapy, CBT alone may be a cost-effective option depending on the quality of life gained per response and its cost effectiveness. From the perspective of the NHS, treating OCD as soon as it is identified, and through cost-effective means, is likely to be better value for money than treating a more severe presentation downstream.