Our study suggests that treating to the target of remission is the preferred strategy over usual care in early RA. After two years of treatment, T2T is cost-effective as it comes with higher costs but also with substantially higher effectiveness. In the T2T group DAS28 remission had been achieved more frequently and there was a larger gain in health-related QALYs compared with UC. The ICUR lies far below the threshold of €80,000 per QALY, which is considered to be an acceptable willingness to pay for one QALY in The Netherlands
. Moreover, the costs to bring one more patient in remission also seem to be acceptable. Results of an extended follow-up analysis of three year data were clearly in favor of T2T, with 64% chance of the T2T strategy coming at lower costs with higher effectiveness compared to UC. To our knowledge, this is the first health economic evaluation of comparing T2T with UC using real-life data.
The drivers of absolute costs and cost differences between T2T and UC were anti-TNF therapy and hospitalization. Our previous studies demonstrated that the majority of the T2T patients achieved remission with conventional DMARDs
[7, 9]. According to the treatment protocol, anti-TNF was prescribed only for a minority of patients whose disease activity remained moderate to high after insufficient effect of conventional DMARDs, thereby preventing overtreatment with anti-TNF agents with their costs and side effects. In the UC group, anti-TNF was initiated later in the disease course, and, therefore, it might be less effective and longer required in patients. Costs due to hospitalization were directly related to RA. The higher number of hospital admissions in the UC group might be explained by less efficient disease control. Data on intra-articular injections were not available. However, we do not think this can explain the difference in costs between T2T and UC, because of the low costs of injections.
The principle of T2T is to aim at achieving and sustaining remission as early as possible. Our expectation is that the extra effort and time spent in the first years of the disease, ultimately result in a reduction of the number of consultations later in the disease course and the possibility of tapering and discontinuing medication in case of sustained remission, thereby diminishing costs. Therefore, we expect that on the long-term, cost savings associated with T2T will increase. Furthermore, better and earlier disease control might lead to more work participation on the long-term, which will ultimately lower the costs of T2T for society and improve quality of life for the patients.
An important strength of this study is the quasi-experimental design containing real-life observational data regarding effectiveness, health-related quality of life and costs of T2T compared with usual care. Moreover, in the DREAM registry all consecutive patients are prospectively followed and a standardized data set is collected. This is in contrast to many health economic evaluations, which often use modeling techniques with many underlying assumptions or use clinical trial data of highly selected patients.
However, this study has some limitations also. Obviously, since our patients are unselected, randomisation between the comparing cohorts was not performed with the risk of confounding by indication present in the comparison. However, participation in either one of the cohorts was determined by living area while patients attended comparable rheumatology clinics, all participating in the DREAM registry and working within the same health care system. We assume that this has limited the possibility of confounding by indication. Furthermore, no relevant differences in baseline characteristics which are prognostic for the treatment effect were found. Second, it should be noted that UC patients were included from 2000 until present, whereas T2T patients were included from 2006 until present. Even though the same treatment options for both groups were available during observation and anti-TNF guidelines have not been changed since 2000, one can assume that UC has changed between 2000 and 2006. A sub-analysis omitting UC patients recruited prior to 2006 showed comparable ICERs, however with less statistical power, leading to the same conclusions. Therefore, this study provides the best possible comparison currently available. A third limitation is that a preference based health-related quality of life measure was not available in the UC cohort, and, therefore, we estimated utilities from the HAQ. The HAQ has been shown to be highly correlated with health state utility values, which are used to calculate QALYs
. Nevertheless, HAQ-derived utilities will only capture change in quality of life generated by the patient’s functional status and not by other factors. We expect that T2T patients improve at more dimensions of quality of life than only function status. Therefore, we believe that this was a conservative analysis. We acknowledge that the use of a mapping method will always be suboptimal to primary collection of utility data. Fourth, we applied a health care perspective, thereby taking into account only direct medical costs. However, the economic burden of RA goes beyond health care costs
[27–29] and a societal perspective would be preferable. RA leads to substantial losses in terms of work productivity which increases with disease duration
[30, 31]. Unfortunately, data on work participation were not available. According to our view, we performed a conservative cost analysis and our expectations are that T2T, which decreases disease activity rapidly and early in the disease course, will have an additional positive effect on non-medical costs (e.g. work productivity, informal care, and paid housekeeping).