Cost-effectiveness analysis of (accelerated) pre-operative versus (conventional) post-operative radiotherapy for patients with oral cavity cancer in Sweden

The incidence of oral cavity cancer (OCC) in Sweden is 11.3 per 100,000 individuals [1], but the malignancy is more common globally [2]. Treatment for resectable OCC often includes combinations of surgery and radiotherapy (RT), but there is no conclusive information on the preferred treatment order [3, 4]. This issue was partly addressed in the multicentre ARTSCAN 2 randomised controlled trial (RCT) [5].

In the ARTSCAN 2 RCT, patients with resectable OCC were randomised to either pre-operative accelerated RT or post-operative conventional RT. The primary outcome was locoregional control (LRC) and secondary objectives included overall survival (OS). LRC was similar for the two treatment regimens [5]. A trend towards better OS for post-operative RT (cf. pre-operative RT) was observed but failed to reach statistical significance. Median and minimum follow-up times for OS were nine and five years, respectively [5].

Apart from survival and morbidity of OCC, the societal cost is also an important consideration for its management [6]. Such costs may be described as direct (i.e., health care resources spent for work-up and treatment) or indirect (i.e., production lost due to sick-leave and early retirement). Furthermore, a cost-effectiveness analysis (CEA) may be performed to ascertain that societal resources are used efficiently [7]. Costs and cost-effectiveness per order of surgery and radiotherapy for OCC have not previously been described.

Information on outcome, morbidity, and societal costs is central for decision-makers when implementing new healthcare routines in clinical practise. The aim of this study was to assess costs and cost-effectiveness for patients with OCC and compare pre-operative with post-operative RT according to the ARTSCAN 2 RCT.

Out of the 240 patients eligible for intention-to-treat analysis in the ARTSCAN 2 RCT [5], 31 individuals were excluded. The reasons for withdrawal are indicated in Fig. 1. Two patients were offered to switch from pre- to post-operative RT and one from post- to pre-operative RT due to temporarily extended waiting times to start of intended treatment. Accordingly, 209 patients were available for the present analysis, i.e., 98 in the pre-operative RT group and 111 in the post-operative RT group.

Data on demographics for the study population are presented in Table 1. The characteristics were well-balanced between the two treatment regimens (Table 1).

There was no difference in LRC between the treatment regimens in the present patient cohort: LRC at 5 years was 77% (95% CI 68–86) for pre-operative RT and 76% (95% CI 64–81) for post-operative RT (detailed data not shown). Likewise, a trend towards better OS for post-operative RT (cf. pre-operative RT) was observed: log-rank p = 0.23 (Fig. 2). OS, when compared at the fixed time point of five years, was however found to be significantly higher for post-operative RT (72% [95% CI 64–81]) compared to pre-operative RT (58% [95% CI 49–69)]) (p = 0.036). The mean life years (LY) lost during the first 5 years was 1.3 for pre-operative RT and 1.0 for post-operative RT, resulting in a mean 0.3 LY gain in favour of post-operative RT.

Direct and indirect costs are presented in Table 2. The total direct cost for work-up and treatment was less for post-operative RT (€ 39,841) compared to pre-operative RT (€ 47,377), and this difference reached statistical significance. Similarly, the subsets of direct costs representing outpatient and inpatient care were less for post-operative RT cf. pre-operative RT, with both subsets reaching statistical significance. The indirect costs for sick-leave and early retirement were similar between the groups.

The intervention, i.e., pre-operative RT, led to significantly higher costs and a lower probability of survival at five years (− 14-percentage points). On the CE-plane, 88% of the pairs were in the northwest quadrant, indicating that pre-operative RT was more costly and less effective, while 12% were in the southwest quadrant (less costly and less effective) (Fig. 3). Thus, pre-operative RT was dominated by post-operative RT. In the sensitivity analysis, the cost-effectiveness results were robust to variations in costs and five-year survival rate. All subgroup analyses were in keeping with the base-case analysis (Table 3).

An overall analysis of the ARTSCAN 2 RCT indicates that pre- and post-operative RT are equally effective alternatives for resectable OCC regarding LRC and OS when combinations of surgery and RT are considered [5]. In addition, the present analysis based on this trial now demonstrates that pre-operative RT is more expensive and less cost-effective compared to post-operative RT.

In this study, the direct cost for pre-operative RT was significantly greater compared to post-operative RT, and it was driven by costs for inpatient as well as outpatient care. The “bottom-up” analysis of direct costs was of importance as it enabled us to clear the data from costs produced by conditions other than, or considered not associated with, OCC and its treatment. As this detailed level of information was available only for the population treated at Skåne University Hospital, these data were imputed for the remainder of the study population. This procedure was validated by comparing the cost-effectiveness  outcome between the patients treated at Skåne University Hospital and the entire study population (data not shown): our conclusions were not affected. A more detailed specification of direct costs can unfortunately not be provided, as the recordings were not uniform over the study period. However, we may speculate that the greater pre-operative RT-group costs for in- and outpatient care might reflect a more marked acute toxicity/morbidity produced by accelerated RT cf. conventionally fractionated RT (i.e., the post-operative treatment), which has been described in the context of head and neck cancer [15].

The present study assessed cost-effectiveness, and one-way sensitivity- and subgroup analyses were performed. In the CEA, uncertainty regarding incremental costs and effects was assessed using non-parametric bootstrapping, which is a method commonly used to describe the distribution of possible mean values [7]: it uses resampling from the data to generate new estimates of the sampling distribution. The outcome indicated that post-operative RT was the preferred/cost-effective treatment: the costs were lower and there was a difference in OS at five years between the regimens in favour of post-operative RT. However, we acknowledge that specific data on quality of life are warranted for the CEA to be fully comprehensive. Cost-effectiveness for specific patient subgroups in terms of age, gender, site, and clinical stage was also assessed, indicating that post-operative RT was even more cost-effective (cf. pre-operative RT) for persons over 65 years of age, men, and tongue/floor of mouth subsite. However, the subgroup results should be interpreted with caution due to small sample sizes and that stratification was not performed for age and gender.

Several studies have estimated societal costs for head and neck cancer, but few exclusively on OCC and, to the best of our knowledge, none with a focus on costs and cost-effectiveness of pre-operative RT versus post-operative RT [16, 17]. Furthermore, in general, previous cost studies on head and neck cancer have key drawbacks. First, they report only direct or indirect costs [18‐24]. Second, they are “top-down” analyses of registry data, not curated against medical records [23, 24]. In contrast, the present study assesses costs specifically for OCC, including direct medical costs of work-up and treatment from a “bottom-up” perspective and indirect costs (e.g., costs due to sick-leave and early retirement), both cleared from costs of co-existing conditions. The advantage of having patient-specific data on societal costs and outcomes is that these can be further used in CEAs, such as in this study, to determine whether a treatment is justified in terms of health gains for patients with OCC.

Two previous studies focus on the impact of the order of surgery and RT in head and neck cancer, both arguing for post-operative RT [25, 26]. However, ARTSCAN 2 is the first RCT that can be analysed in this context, and the interpretation is that the regimens are equal in terms of LRC and OS [5]. In that context, the present observations on costs and cost-effectiveness may be of particular interest to surgeons/oncologists and other decision-makers when choosing the treatment order for resectable OCC: we forward a treatment order of surgery followed by RT. However, a complete analysis of complications, morbidities, and quality of life associated with OCC and its treatment is warranted and will be analysed later by the ARTSCAN 2 Study Group. Since the direct cost data obtained in this study are from a public, non-profit health care system, the cost level is likely generalisable to similar systems, e.g., the Nordic Countries.

In contrast, extrapolation to private health care and government-run insurance programmes should be avoided. The results of this study will reinforce the routine at hospitals where post-operative RT is currently standard. If the opposite is the situation, the current results, in connection with other publications of the ARTSCAN 2 study, suggest that a change of treatment order may be considered.

Based on data from the ARTSCAN 2 RCT, we conclude that post-operative RT for resectable OCC is the dominant alternative compared to pre-operative RT.