Cost-effectiveness analysis comparing in-office posterior nasal nerve ablation to surgical therapies

Abstract

Background:

Recently, in-office posterior nasal nerve ablation (PNA) devices have offered a new tool to treat refractory chronic rhinitis, but their cost-effectiveness relative to traditional interventions such as vidian neurectomy (VN) and posterior nasal neurectomy (PNN) remains unexplored.

Objective:

To compare the cost-effectiveness of these interventions in patients with refractory chronic rhinitis.

Methods:

A decision tree with embedded Markov models was created to compare the cost-effectiveness of PNN, VN, and PNA, measured in quality-adjusted life years (QALYs) over a 30-year time horizon with a $100,000/QALY willingness-to-pay threshold. One- and two-way sensitivity analyses were completed.

Results:

Sensitivity analysis found that in-office PNA became cost-effective compared to VN when patients undergoing PNA were less than 20% more likely than VN to have symptoms recur; this value was assumed to be twice as likely in the base case. In the base case, however, VN and in-office PNA were more effective and less expensive than PNN, while VN was cost-effective when compared to in-office PNA (incremental cost-effectiveness ratio $11,616.24/QALY). Other assumptions were not found to considerably impact incremental cost-effectiveness.

Conclusion:

Although highly limited by currently available data, PNA may be cost-effective compared to VN as long-term outcomes on the durability of its effects emerge. These data should not be used by payers considering coverage or utilization since long-term data is still nascent. However, that as new technologies emerge for rhinitis, it will be important to monitor longer-term outcomes to identify high value care, but based on limited data PNA devices may meet this standard.

1. Introduction

Chronic rhinitis is a common rhinologic condition, affecting 10% to 40% of the population, often with considerable quality of life implications.¹ It is a frequent reason for office visits, associated with annual costs upwards of $3 billion.² Previous studies have assessed a variety of treatment options for clinical efficacy and cost effectiveness, including topical medical therapy, systemic treatments, and immunotherapy.³ However, many cases of allergic and non-allergic rhinitis are refractory to medical therapy; in these patients, procedural-based and surgical approaches to rhinitis have been explored.

Traditional surgical approaches to intractable chronic rhinitis include vidian neurectomy (VN) and, more recently, posterior nasal neurectomy (PNN).⁴ However, these procedures are infrequently pursued by patients because they require general anesthesia and present unique potential complications including dry eye, post-operative bleeding, and numbness.⁵ More recently, posterior nasal nerve ablation (PNA) devices have offered a new and accessible treatment option to manage refractory chronic rhinitis, as an in-office procedure under local anesthesia.^6,7 PNA devices may also potentially mitigate the side effects associated with traditional surgical methods.⁸ However, as these new tools are adopted, cost and efficacy of these single-use, disposable PNA devices will need to be carefully considered.

While VN, PNN, and PNA have been shown to improve symptoms of chronic rhinitis in some patients, a cost-effectiveness analysis comparing these approaches has not yet been performed. This information would be critical to providing value-based care, optimizing resource utilization, and guiding further development and evaluation of in-office devices for chronic rhinitis management. Thus, this study sought to compare the cost-effectiveness of VN, PNN, and PNA in patients with chronic rhinitis refractory to medical management.

2. Methods

2.1. Decision tree and Markov model

The base case for this economic analysis was a patient with chronic rhinitis refractory to medical therapy and seeking a surgical or procedural alternative. The base case patient in this model was assumed to be 53 years old, which was the average age of patients in the first major PNA study.⁶ A decision tree and three Markov models were created to model patients who chose in-office PNA, PNN, or VN. The primary outcome was the cost-effectiveness of the interventions relative to each other, measured as an incremental cost-effectiveness ratio (ICER) in quality-adjusted life years (QALYs) with a $100,000/QALY cost-effectiveness threshold.

Patients moved through the decision tree prior to encountering the terminal Markov models (Figure A.1). The decision tree included major intraoperative complications (i.e., anesthesia-related) and postoperative/post-procedural complications (e.g., bleeding, infection, or dry eye). Patients then entered a multi-state Markov model unique to their choice of procedure, where states included permutations of dry eye, numbness, rhinitis symptoms, and death (Figure A.2). Those who no longer had rhinitis symptoms did not require further rhinologic care, while those who did were modeled to have annual rhinologist visits with nasal endoscopy. Patients were assumed not to seek further procedural interventions for their rhinitis.

This analysis was performed from the perspective of the U.S. government payer with a 30-year time horizon and 3-month cycle length. A 3% discount rate for costs and QALYs was used and half-cycle corrections were included.

2.2. Effectiveness and complications

Effectiveness was modeled from several effectiveness studies, albeit with a limited amount of published data (Table 1). Initial probabilities of treatment success and complications after each treatment modality were modeled using values from prospective and retrospective studies reported in the literature. Vision loss after VN, while theoretically possible, was not reported commonly in the literature and was thus not included in the model. These probabilities were modeled as joint probabilities when assigned as initial probabilities for the Markov model. Transition probabilities in the Markov model were modeled using rates of post-procedural numbness and dry eye resolution. No long-term data was found reporting rates of chronic rhinitis symptoms returning years post-procedure for any of these procedures, and so was modeled as 1% per quarter for VN and PNN. Another variable modeled the relative rate of chronic rhinitis symptoms returning for PNA (3% per quarter in the base case) and was later evaluated with a one-way sensitivity analysis.

Table 1.

Model parameters of probability and health utility values with their sources.

Variable	Value	Source
Probabilities
Probability of bleeding after PNA	0.006	Balai et al.;27 Stolovitsky et al.;28 Del Signore et al.;29 Ow et al.;22 Chang et al.7
Probability of bleeding after PNN	0.007	Kikawada et al.4
Probability of bleeding after VN	0.000	Tan et al.30
Probability of major bleeding requiring OR management after PNA (given patient has bleeding)	0.500	Chang et al.7
Probability of major bleeding requiring OR management after PNN (given patient has bleeding)	1.000	Kikawada et al.4
Probability of major bleeding requiring OR management after VN (given patient has bleeding)	0.000	Tan et al.30
Probability of all-cause death	-	SSA actuarial tables9
Probability of dry eye after PNA	0.009	Balai et al.;27 Stolovitsky et al.;28 Del Signore et al.29 Chang et al.7
Probability of dry eye after PNA resolving if it develops	0.000	Chang et al.7
Probability of dry eye after VN resolving within 3 months if it develops	1.000	Tan et al.30
Probability of sinusitis after PNA	0.018	Chang et al.7
Probability of major intraoperative complications in surgery	0.010	Scangas et al.18
Initial probability of no rhinitis after PNN	0.391	Takahara et al. 31
Initial probability of no rhinitis after VN	0.647	Tan et al. 30
Initial probability of no rhinitis after PNA	0.702	Balai et al.;27 Stolovitsky et al.;28 Del Signore et al.29
Initial probability of no rhinitis and dry eye after PNA	0.005	Balai et al.;27 Stolovitsky et al.;28 Del Signore et al.;29 Chang et al.7
Initial probability of no rhinitis and no dry eye after PNA	0.697	Balai et al.;27 Stolovitsky et al.;28 Del Signore et al.;29 Chang et al.7
Initial probability of rhinitis and dry eye after PNA	0.002	Balai et al.;27 Stolovitsky et al.;28 Del Signore et al.;29 Chang et al.7
Initial probability of rhinitis and no dry eye after PNA	0.296	Balai et al.;27 Stolovitsky et al.;28 Del Signore et al.;29 Chang et al.7
Probability of rhinitis returning after cure with PNA (per 3 month Markov cycle)	0.030	No data, assumption tested with sensitivity analysis
Initial probability of no rhinitis and no numbness after PNN	0.391	Takahara et al.31
Initial probability of rhinitis and no numbness after PNN	0.609	Takahara et al.31
Initial probability of no rhinitis with numbness after PNN	0.000	No literature reporting facial numbness after PNN
Probability of rhinitis returning after cure with PNN (per 3 month Markov cycle)	0.010	No data, assumption tested with sensitivity analysis
Initial probability of no rhinitis and no numbness after VN	0.586	Tan et al.30
Initial probability of no rhinitis with numbness after VN	0.061	Tan et al.30
Initial probability of rhinitis with no numbness after VN	0.320	Tan et al.30
Initial probability of rhinitis with numbness after VN	0.033	Tan et al.30
Probability of rhinitis returning after cure with VN (per 3 month Markov cycle)	0.010	No data, assumption tested with sensitivity analysis
Probability of facial numbness resolving after VN	Modeled by a stepwise function as resolving by 1 year as seen in Tan et al.	Tan et al.30
Disutilities
Disutility of moderate rhinitis	0.085	Retzler et al.11
Disutility of moderate dry eye	0.100	Schiffman et al.14
Disutility of facial numbness	0.050	No data, assumption tested with sensitivity analysis
Disutility of general anesthesia/operating room complications (2 weeks calculated over 1 year)	0.000	Selai and Rosser,12 Tengs and Wallace13
Disutility of infection (assuming 2 weeks of SNOT-22 of 40)	0.023	Crump et al.15

Quarterly mortality rates were cycle-corrected from Social Security Administration actuarial tables⁹ and weighted by the ratio of male:female patients in the first PNA prospective study.⁶

Quality of life was represented using average health utility values (HUVs) abstracted from the literature (Table 1). Quarter-cycle corrections were made to all HUVs incorporated in the model, but reported values are annualized for clarity. Baseline HUVs by age were represented by average EuroQol-5D index of U.S. adults,¹⁰ and the effects on quality of life of complications or of chronic rhinitis were modeled as disutilities. Rhinitis quality of life was modeled as a disutility of “moderate” rhinitis from data abstracted from a standard gamble study of adults with rhinitis.¹¹ The quality of life impact of a major operating room complication was modeled using HUVs representing acute illness in a hospital, assuming an average 2 weeks of impacted quality of life.^12,13 The HUV decrement for dry eye was abstracted from a utility study of patients with dry eye disease.¹⁴ Post-procedural sinusitis was modeled as a disutility that assumed 2 weeks with a SNOT-22 score of 40.¹⁵ No studies were found in the literature reporting HUVs for facial numbness, so the disutility was assumed to be 0.05 and a one-way sensitivity analysis was later conducted on this value.

2.3. Costs

Costs were modeled from the perspective of the U.S. government payer (Table 2). Procedure and surgery costs were calculated using the CY 2022 CMS Physician Fee Schedule¹⁶ using professional fee CPT^® codes reported by three academic rhinologists at different facilities. The Medicare geographic practice cost index was assumed to be 1.00 and non-facility reimbursement rates were used. PNA was coded as 31231 (diagnostic nasal endoscopy) and 30117 (removal of intranasal lesion) with a −51 bilaterality modifier and 99213 (outpatient office visit, 20–29 minutes). Two follow-up visits were assumed, each with a nasal endoscopy and 99213 (outpatient office visit, 20–29 minutes) for a total of $2,635.13.

Table 2.

Model cost parameters with their sources.

Variable	Value	Source
Initial cost of PNA (procedure and clinic follow-up)	$2,635.13	See methods section
Initial cost of PNN (procedure and clinic follow-up)	$4,414.24	See methods section
Initial cost of VN (procedure and clinic follow-up)	$4,793.42	See methods section
Cost of major operating room complication	$16,877.00	Rudmik et al.,32 Scangas et al.18
Cost of postprocedural infection	$559.00	Rudmik et al.,32 Scangas et al.18
Cost of minor epistaxis managed conservatively	$400.00	Rudmik et al.,32 Scangas et al.18

Both posterior nasal neurectomy and vidian neurectomy were modeled as outpatient surgeries. Vidian neurectomy was assumed to need only an overnight observation stay (no additional charge) and was not billed as an inpatient case. To ensure anesthesia and facility fees were included, costs were abstracted from a study¹⁷ estimating cost by component procedures of endoscopic sinus surgery (ESS). PNN is often billed using sphenopalatine artery ligation codes, and, with a work RVU of 8, is most similar to maxillary antrostomy and anterior ethmoidectomy. After including two follow-up visits with nasal endoscopy, this was estimated to cost $4,414.24. VN, billed using the pterygomaxillary fossa surgery CPT code, was most similar in work RVU (9.8) to frontal sinus and anterior ethmoid surgery, and was estimated after including two follow-up visits with endoscopy to cost $4,793.42. Costs of major operating room complications, major and minor epistaxis, and infections were taken from the literature.¹⁸

All costs were adjusted to 2021 US dollars using the Consumer Price Index for medical care.¹⁹

2.4. Sensitivity analysis

One-way sensitivity analyses were done for all variables with the primary outcome and evaluated with a tornado diagram. Probabilistic sensitivity analysis was not possible due to data limitations that did not provide information on variance for all variables abstracted from the literature.

This was a cost-effectiveness analysis reported using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) guidelines.²⁰ Analysis was performed using TreeAge Pro Healthcare Version 2022 (TreeAge Software, Inc., Williamstown, Massachusetts) and Microsoft Excel (Microsoft Corp., Redmond, Washington). Institutional review board approval was not required for this project as there were no human subjects data used.

3. Results

3.1. Cost-effectiveness

Using a 30-year time horizon, the model showed patients choosing in-office posterior nasal nerve ablation incurring $6,745.16 of cost with 13.62 QALYs. Posterior nasal neurectomy patients incurred $9,127.83 of cost with 13.53 QALYs. Vidian neurectomy patients incurred $8,464.46 with 13.77 QALYs. Posterior nasal neurectomy was dominated in cost-effectiveness by both in-office posterior nasal nerve ablation and vidian neurectomy. Vidian neurectomy was more expensive than in-office posterior nasal nerve ablation but generated more QALYs with an ICER of $11,616.24/QALY relative to PNA, below the a priori $100,000/QALY threshold (Figure 1).

Figure 1.

Cost effectiveness plot of in-office posterior nasal nerve ablation, posterior nasal neurectomy, or vidian neurectomy.

3.2. One-way sensitivity analysis of the rate of chronic rhinitis symptoms returning

A one-way sensitivity analysis was completed on the relative probability of rhinitis returning after cure between PNA and PNN/VN. With relative probability values ranging from 0 to 3.00 (assumed to be 2.00 in the base case), we found that in-office PNA became cost-effective compared to VN when the risk of rhinitis symptoms recurring post-treatment in PNA compared to VN fell below 1.20 and the ICER comparing PNA to VN crossed the a priori $100,000/QALY threshold when the relative risk fell below 1.29 (Figure 2).

Figure 2.

One-way sensitivity analysis of the rate of chronic rhinitis symptoms returning.

3.3. Other one-way sensitivity analyses

Other one-way sensitivity analyses were completed, including the HUV of facial numbness, HUV of moderate rhinitis, probability of dry eye after VN, cost of PNA, and probabilities of post-operative/procedural bleeding (Appendix). These variables were not found to considerably impact incremental cost-effectiveness.

4. Discussion

This study sought to investigate the cost-effectiveness of existing surgical interventions for chronic rhinitis, including VN and PNN, as compared to in-office PNA, in patients with chronic rhinitis refractory to medical management. Our results, in the setting of limited long-term data, indicate that VN and in-office PNA both were more effective and less expensive than PNN. VN was cost-effective when compared to in-office PNA at the $100,000/QALY willingness-to-pay threshold. However, a later sensitivity analysis found that in-office PNA became cost-effective compared to VN when the risk of rhinitis symptoms recurring post-treatment in PNA compared to VN fell below 1.20; this value was assumed to be 2.00 in the base case model. Other outcomes, such as the probability of dry eye after VN, cost of PNA, and probabilities of post-operative/procedural bleeding, were not found to considerably impact incremental cost-effectiveness in sensitivity analyses.

While published evidence^3,21 has investigated the cost-effectiveness of various treatments for chronic rhinitis, no group has yet evaluated surgical and procedural options for the condition. The presented economic evaluation demonstrated that VN and in-office PNA appear to be more effective and less expensive than PNN. However, VN was cost-effective when compared to in-office PNA under the assumption that the relative risk of symptom recurrence post-treatment in PNA compared to VN was 2.00, but this was not the case when this relative risk of symptom recurrence fell below 1.20. Notably, there is not yet long-term data demonstrating the comparative efficacy between the two treatment options given the recent emergence of PNA, though recent evidence suggests the longevity of PNA treatment response to be considerable, with over 80% of patients showing the minimal clinically important difference for symptom improvement at up to 24 months in a 2021 study by Ow et al.²² Our work could set a rudimentary clinical performance benchmark that medical technology developers could seek to achieve with further research into and advancement of in-office PNA devices to render their device the most cost-effective treatment option.

Interestingly, several variables in the model, the HUV of facial numbness, HUV of moderate rhinitis, probability of dry eye after VN, cost of PNA, and probabilities of post-operative/procedural bleeding were not found to significantly impact incremental cost-effectiveness in one-way sensitivity analyses. Additionally, the cost of the PNA was likely not an impactful variable in the sensitivity analysis since it was considerably lower than the cost of VN and PNN.

This study has several inherent limitations that warrant discussion. First, the paucity of available long-term data resulted in a lack of variance estimates for many of the variables abstracted from the literature, which limited probabilistic sensitivity analysis (PSA). While variance could be estimated by using the sample size of the probability estimate and assuming a binomial distribution, it was felt that this level of assumption may overstate the quality of long-term data available for this study. This limitation, notably, was far more pronounced in the scanter data available for VN and PNN; more recent PNA studies are general well-designed prospective cohort studies or RCTs. Second, data on probabilities of complications following VN or PNN and disutility values were very limited. Critically, there was no long-term data for any of the procedures on the probability of symptom recurrence years post-procedure—this therefore had to be evaluated with one-way sensitivity analyses. Cost data is also highly variable between institutions, as billing for VN and PNN may use unlisted codes with significant cost implications. Billing for PNA devices is also variable and challenging to estimate. We considered each of these challenges and settled on our approach after discussing billing practices with several high-volume fellowship-trained rhinologists. The limited available data was also reflected in assumptions made in our long-term projections. Markov transition probabilities in each surgical cohort were presumed to correctly represent long-term outcomes in the treatment of chronic rhinitis, an assumption that has not yet been reliably shown for chronic rhinitis but has been for ESS.^23–25 A linear relationship was assumed between HUVs and average cost of yearly care based on previous cost-effectiveness analyses.^25,26 Third, in this analysis, we grouped under “PNA” devices using both radiofrequency energy and cryoablation. As more long-term data becomes available on these methods for PNA, further work can compare the cost-effectiveness of each. Fourth, in the long-term we did not factor in further PNA treatments, but some patients will undergo this procedure multiple times. Finally, in this study, we considered the multifaceted condition of chronic rhinitis to be a single disease; however, allergic and non-allergic rhinitis are clinically and physiologically different conditions.

Despite these limitations, this is the first study to compare the cost-effectiveness of various surgical strategies for the treatment of chronic rhinitis and is also the first study to quantify the cost-effectiveness of newer in-office PNA techniques.

5. Conclusions

To our knowledge, this is the first study to compare the cost-effectiveness of various surgical interventions for chronic rhinitis and also the first cost-effectiveness analysis of PNA. We find that in-office PNA may be cost-effective when compared to VN if the relative risk of symptom recurrence is below 1.20 when compared to VN. In-office PNA’s relative cost-effectiveness to VN is thus highly dependent on the long-term durability of its treatment, but the most recent long-term data on 2-year outcomes suggest it may be long-lasting. This analysis, based on early and limited data, lays some groundwork upon which further economic analyses can be conducted as in-office PNA technology advances and more long-term data on the stability of its effects is published. These results should not be used by payers or administrators considering coverage or utilization given the early stage of research and current dearth of long-term data in the literature. As emerging technologies enter this space, it will be important to monitor longer term outcomes to identify high-value, cost-effective care, but current limited data suggest that in-office PNA devices could meet this high standard.