Cost of ciprofloxacin/dexamethasone vs. amoxicillin/clavulanic acid for the treatment of acute otitis media in tympanostomy tube patients in the US

Summary

A deterministic decision-analytic model estimated the cost effectiveness of two therapeutic alternatives for the treatment of acute otitis media in tympanostomy tube patients (AOMT).

Three tiers of AOMT antibiotic therapy were modelled, with successive tiers representing re-treatment of failures from the preceding tier. First-tier therapy compared ciprofloxacin/dexamethasone (CD) ear drops with oral amoxicillin/clavulanic acid (ACA) using clinical trial data. ACA was used as second-tier therapy for re-treatment of initial CD failures, whilst CD was used to re-treat initial ACA failures. Cefdinir was used as third-tier therapy. Costs were presented from a US payer's perspective. Sensitivity analysis established the robustness of the model.

The expected cost per cure was $219 in the CD pathway vs. $293 in the ACA pathway. Given its effectiveness and side-effect profile, CD was more cost effective than ACA for AOMT therapy. CD dominated ACA despite the higher acquisition cost of CD.

Keywords: :

• antibacterial agents
• anti-infective agents
• anti-inflammatory agents
• costs
• cost analysis
• tympanostomy

Introduction

Acute otitis media (AOM) is the most common bacterial infection in paediatric patients and is the leading indication for outpatient antimicrobial use1. Seventy-five percent of children experience at least one episode of otitis media and almost 50% will have three or more ear infections by their third birthday. Medical costs and lost wages due to otitis media are estimated at $5 billion in the US annually2.

Children who have experienced repeated episodes of AOM may be considered candidates for insertion of tympanostomy tubes3–5. Tympanostomy tube placement is the second most common surgery in children in the US5,6, with more than 1 million children receiving tympanostomy tubes each year7. Following insertion of the tubes, some patients continue to experience drainage secondary to bacterial infection, referred to as acute otitis media in tympanostomy tube patients (AOMT)8,9. Ciprodex®* (ciprofloxacin 0.3% and dexamethasone 0.1%) Sterile Otic Suspension (CD) is approved by the US Food and Drug Administration for the treatment of AOMT and acute otitis externa. A number of comparative studies have evaluated the efficacy of this medication10–12. A previously reported economic analysis demonstrated that CD economically dominated ofloxacin ear drops for the treatment of AOMT12.

The objective of this study was to develop an economic model that compared the cost effectiveness of CD ear drops with an oral antibiotic for the treatment of AOMT.

Methods

Economic model structure

An economic model was developed to calculate the costs and outcomes of AOMT treatment initiated with Ciprodex® (CD) or Augmentin®† ES600 (600 mg amoxicillin and 42.9 mg clavulanic acid; ACA). Each successive therapeutic tier of the clinical pathway used a different antibiotic than the previous therapeutic choice (i.e. not a repeat course of therapy). CD and ACA were selected as first-tier AOMT therapies in order to compare the oral antibiotic of choice (ACA), as identified in a physician practice pattern survey, with the most potent ototopical with an approved indication for AOMT (CD)11. The antibiotics selected for the clinical algorithm were based on a survey of 300 practicing physicians conducted in Autumn 2006. ACA was selected as second-tier therapy following treatment with CD, whilst CD was selected as second-tier therapy following treatment with ACA. The drug of choice for third-tier therapy was cefdinir regardless of prior antibiotic selection.

Using a decision-analytic approach, a cohort of patients (n=1000 in each initial pathway) was modelled through the AOMT clinical algorithm for up to three tiers of therapy. Figure 1 depicts the AOMT clinical algorithm along with the effectiveness probabilities for each antibiotic. Primary outcomes for the model include cumulative success rates and total treatment costs for the respective pathways. In addition, the model calculates the percentage of patients who succeed at each tier of therapy and the distribution of costs by type of resource (physician, antibiotic, side-effect treatment, laboratory/procedure).

Figure 1. Decision analytic clinical algorithm.

AOMT patients were initially treated with either ototopical CD (four drops in the affected ear twice daily for 7 days) or systemic ACA (45 mg/kg/day in divided doses for 10 days) as first-tier therapy. Following treatment, patients achieved clinical success or failure; failure could have resulted from lack of efficacy or discontinuation resulting from side effects. Patients who achieved clinical success could also have experienced side effects. Patients who did not achieve clinical success were assumed to receive subsequent physician visits for evaluation as well as prescriptions for subsequent tiers of therapy. The clinical algorithm followed patients until cure or through three tiers of therapy, whichever came first.

Data sources and parameters

For first-tier therapy, clinical parameters were taken from a recently published clinical trial13. This clinical trial provided the effectiveness rates as well as both the frequency and type of side effects used in the first tier of the clinical algorithm. For second- and third-tier therapy, the clinical parameters were taken from a physician survey and prescribing information. Table 1 summarises the clinical input parameters for the model.

Table 1. Rates of clinical cure and treatment discontinuation due to side effects*.

Therapy	Cure rate (total) (%)	Cure rate with side effect (%)	Failure rate without side effect (%)	Failure rate with side effect (%)	Source(s)
First-tier therapy
Ototopical CD	84.6	12.8	12.8	2.6	Clinical trial
Systemic ACA	58.5	29.3	39.0	2.4	Clinical trial
Second-tier therapy
Ototopical CD	89.0	6.8	8.0	3.0	Physician survey, prescribing information
Systemic ACA	87.4	19.1	9.6	3.0	Physician survey, prescribing information
Third-tier therapy
Cefdinir	90.0	13.6	8.0	2.0	Physician survey, prescribing information

CD, ciprofl oxacin/dexamethasone; ACA, amoxicillin/clavulanic acid.

*Numbers may not sum to 100% owing to rounding.

Costs associated with physician visits, procedures and laboratory tests were taken from the 2006 Physician Fee and Coding Guide14. Drug costs were based on average wholesale prices and taken from the 2006 Red Book15. Table 2 summarises the resource and cost input parameters for the model-weighted costs for laboratory tests and side effects. Typical patterns of laboratory and procedure use and side effects associated with each line of therapy were valued using weighted averages based on the frequency of mention in the physician survey.

Table 2. Costs.

Resource	Source/description	Cost ($)
Medications to treat AOMT	Red Book, AWP^15
Ciprofloxacin/dexamethasone (ototopical)	0.3–0.1%, 7.5 ml (NDC 00065-8533-02)	94.44
Systemic amoxicillin/clavulanic acid (PO)	400 mg/ml, 100 ml (NDC 00029-6092-51)	87.85
Cefdinir (PO)	250 mg/5 ml, 100 ml (NDC 54569-5650-00)	152.49
Physician contacts	Medical Fee and Coding Guide^14
Physician visit, non-specialist	CPT 99212	56
Physician visit, specialist	CPT 99252	150
Urgent care	CPT 99283	175
Physician phone call	CPT 99372	27
Laboratories and procedures	Medical Fee and Coding Guide^14
Irrigation/cleaning	CPT 69210	77
Specimen culture	CPT 87070	37
Kirby–Bauer sensitivity test	CPT 87184	29
Stool panel	CPT 87046	22
Prescription medications to treat side effects	Red Book, AWP^15
Phenergan	Suppository, 12.5 mg, quantity 4 (NDC 54569-0361-00)	17.49
Tigan Pediatric	Suppository, 2%–100 mg, quantity 5 (NDC 54868-0340-01)	17.33
Reglan	Tablet, 10 mg, quantity 30 (NDC 54868-0513-00)	38.86
Lomotil	Solution, PO, 60 ml (NDC 54868-1511-01)	21.11
OTC medications to treat side effects	Red Book, AWP^15
Diphenhydramine cream	Cream, TP, 2%, quantity 30 g (NDC 45802-0358-03)	2.70
Diphenhydramine	Solution, PO, cherry flavoured, 12.5 mg/5 ml, quantity 118 ml (NDC 61073-0026-04)	4.23
Dramamine	Chewable tablet, 50 mg, quantity 8 (NDC 00009-3648-04)	3.28
Ibuprofen Children's	Suspension, PO, berry flavoured, 100 mg/5 ml, quantity 120 ml (NDC 37205-0646-05)	4.80
Imodium	Liquid, PO, 1 mg/5 ml, quantity 60 ml (NDC 50580-0293-02)	4.36
Kaopectate Children's	Suspension, PO, cherry flavoured, 262 mg/15 ml, quantity 180 ml (NDC 00009-5271-01)	4.27
Loratadine	Tablet, 10 mg, quantity 10 (NDC 49348-0542-01)	4.99
Tums	Chewable tablet, 500 mg, quantity 36 (NDC 00766-0741-62)	1.44
Acidophilus wafer	Wafer, PO, berry flavoured, 100 ml, quantity 100 (NDC 74312-0057-21)	6.09
Lactobacillus	Caplet, quantity 50 (NDC 24038-0100-12)	4.19
Topical steroid (hydrocortisone)	Cream, TP, 1%, quantity 30 g (NDC 00472-0343-56)	3.48
Topical antifungal (Gentian violet)	Solution, TP, 1%, 30 ml (NDC 24385-0003-91)	2.24
Laboratories and procedures	Weighted average based on physician survey and Medical Fee and Coding Guide^14
At initiation of first-tier therapy	7% specimen culture and sensitivity test (CPT 87070/87184), 5% CBC (CPT 85027), 2% ESR (CPT 85651), 2% X-ray (CPT 71034/-26), 2% CT without dye (CPT 70480/-26)	27.91
At initiation of second-tier therapy	22% specimen culture and sensitivity test, 8% CBC, 2% ESR, 3% X-ray, 3% CT without dye	49.39
At initiation of third-tier therapy	43% specimen culture and sensitivity test, 8% CBC, 2% ESR, 3% X-ray, 1% CT without dye	45.45
Side-effect treatment	Weighted average based on physician survey, Red Book^15 and Medical Fee and Coding Guide^14
Diarrhoea/gastrointestinal discomfort	30% OTC medications, 25% dietary change/add supplement, 15% physician phone call or outpatient visit, 20% no additional resources	11.63
Nausea/vomiting	20% physician phone call or outpatient visit, 18% prescription medication, 32% no additional resources	11.19
Ear irritation	29% physician visit, 21% topical steroid, 3% irritation/debridement, 32% no additional resources	4.68
Allergic reaction	27% OTC topical, 25% OTC antihistamine, 22% physician phone call or outpatient visit, 18% no additional resources	11.53

AOMT, acute otitis media in tympanostomy tube patients; AWP, average wholesale price; CPT, Current Procedural Terminology; NDC, National Drug Code; PO, oral; OTC, over the counter; TP, topical; CBC, complete blood count; ESR, erythrocyte sedimentation rate; CT, computed tomography.

Using a weighted average technique, laboratory and procedure costs were developed for each tier of therapy. For example, physicians reported that, at presentation, 7% of patients received a culture and sensitivity test, 5% had a complete blood count (CBC) and 2% had measurement of erythrocyte sedimentation rate, a radiograph or computed tomography without contrast. The CBC cost was $29, but since only 5% of patients had one, the average cost per patient for CBC was $1.45 ($29 × 0.05). This process was used to estimate and sum the average cost per patient for laboratory tests and procedures. A similar process was used to develop the costs associated with side effects. For each type of side effect, a substantial percentage of physicians reported that there would be no additional resources associated with the event, as parents would have been notified about the possibility of the side effects and given instructions in advance. Thus, not all occurrences of side effects had a cost associated with them. Although the rates of side effects were medication specific, it was assumed that side-effect costs were medication independent. For example, whilst the occurrence of nausea was higher for oral antibiotics relative to ototopical therapy, the cost of treating nausea was the same regardless of the therapy or location within the clinical pathway (e.g. first-tier ACA or third-tier cefdinir).

The economic model was constructed using Microsoft Excel®. Only the direct costs associated with AOMT therapy were considered in this analysis. All costs are presented in 2006 US$. As with most acute infections treated in the US, the clinical algorithm assumed that predominantly all AOMT patients were treated until cured. The economic implications are that all direct costs, including the costs associated with re-treatment of therapeutic failures, are considered in the analysis of the overall costs.

Sensitivity analysis

All model parameters were varied to determine the impact on economic outcomes and to establish model robustness. Both univariate and probabilistic sensitivity analyses were performed on this model.

Results

Physician survey

The physician survey provided key elements for developing the AOMT treatment algorithm using information gathered from 300 practicing physicians (50 otolaryngologists, 100 paediatricians and 150 family practitioners). The survey provided information on the AOMT antibiotics of choice and the type, frequency and treatment of side effects associated with the antibiotics. Interestingly, the survey revealed a therapeutic crossover of first-tier therapy for second-tier therapy. Physicians who selected oral ACA as first-tier treatment preferred ototopical CD for second-tier therapy. The reverse was true for the CD pathway; physicians who initiated therapy with ototopical CD preferred ACA for second-tier therapy. Based on the physician survey, the estimated success rates for second-tier ACA and CD therapy were 87.4 and 89.0%, respectively. Lastly, the drug of choice for third-tier therapy was oral cefdinir, regardless of the prior antibiotic used, with an estimated efficacy rate of 90%. Figure 1 illustrates the clinical algorithm with respect to antibiotic selection and efficacy at each tier of therapy.

Information on the type and frequency of side effects associated with first-tier AOMT therapy was taken from the clinical trial, whilst the physician survey established the type and frequency of side effects for second- and third-tier therapies. The most common side effects were localised allergic reaction (5.8%) and ear irritation (5.1%) for CD and diarrhoea (24.4%), nausea/vomiting (7.3%) and allergic reaction (12.2%) for ACA. For cefdinir, side effects were diarrhoea (8%), ear irritation (4.4%) and nausea/vomiting (2.4%).

A weighting approach based on the number of physician mentions of preferred therapeutic options for each class of side effect was used to establish the costs associated with the treatment of side effects. The weighted average costs for each class of side-effect therapy were $10.51 for the treatment of diarrhoea, $11.19 for the treatment of nausea, $11.53 for the treatment of allergic reactions, $4.66 for the treatment of ear irritation and $32.77 for the treatment of ‘other’ side effects. The aggregate side-effect costs for first-tier therapy were $526 and $16,707 for CD and ACA treatment, respectively.

Economic analysis

The economic analysis demonstrated that initiating AOMT therapy with CD resulted in lower costs at every tier of therapy compared with initiating treatment with ACA. The total estimated costs to cure a 1,000-patient cohort were $218,757 for CD therapy versus $293,084 for ACA therapy, a cost saving of $74,327 favouring CD (Table 3). In addition, each tier of therapy in the CD pathway demonstrated cost savings relative to the ACA pathway. Despite substantially higher drug cost, the first-tier costs for a 1,000-patient cohort treated with CD was estimated at $178,876 compared with $188,467 for the ACA cohort, a cost saving of $9,591.

Table 3. Outcomes per tier for 1,000-patient cohorts.

	CD	ACA
First-tier therapy
No. of patients	1,000	1,000
% cured	84.6	58.5
Antibiotic cost ($) (proportion of total cost)	94,440 (52.8%)	87,850 (46.6%)
Physician cost ($) (proportion of total cost)	56,000 (31.3%)	56,000 (29.7%)
Side-effect costs ($) (proportion of total cost)	526 (0.3%)	16,707 (8.9%)
Laboratory/procedure costs ($) (proportion of total cost)	27,910 (15.6%)	27,910 (14.8%)
Total costs ($)	178,876	188,467
Second-tier therapy
No. of patients	154	415
% cured	87.4	89.0
Antibiotic cost ($) (proportion of total cost)	13,515 (39.6%)	39,158 (43.0%)
Physician cost ($) (proportion of total cost)	11,508 (33.7%)	31,015 (34.1%)
Side-effect costs ($) (proportion of total cost)	1,511 (4.4%)	360 (0.4%)
Laboratory/procedure costs ($) (proportion of total cost)	7,598 (22.3%)	20,479 (22.5%)
Total costs ($)	34,133	91,012
Third-tier therapy
No. of patients	19	46
% cured	90	90
Antibiotic cost ($) (proportion of total cost)	2,949 (51.3%)	6,980 (51.3%)
Physician cost ($) (proportion of total cost)	1,810 (31.5%)	4,285 (31.5%)
Side-effect costs ($) (proportion of total cost)	110 (1.9%)	260 (1.9%)
Laboratory/procedure costs ($) (proportion of total cost)	879 (15.3%)	2,081 (15.3%)
Total costs ($)	5,748	13,606
Grand total
Costs ($)	218,757	293,085
Patients cured	998	995
Average cost per cure ($)	219	294

CD, ciprofloxacin/dexamethasone; ACA, amoxicillin/clavulanic acid.

Given the first-tier efficacy rates, 15.4% of the CD cohort proceeded to second-tier therapy compared with 41.5% in the ACA cohort. The substantial difference in the number of patients entering second-tier therapy led to a substantial difference in the relative cost of the alternative pathways. The total cost associated with second-tier therapy for the initial CD pathway was $34,133 compared with $91,012 for the initial ACA pathway, a cost saving of $56,879 favouring initial CD therapy.

Based on the physician survey, the antibiotic choice for third-tier therapy was cefdinir regardless of the respective pathway followed. Consequently, the difference in costs for third-tier therapy is proportional to the number of patients failing second-tier therapy and entering the last tier. The total costs associated with third-tier therapy were $5,748 for patients from the initial CD cohort and $13,606 for patients from the initial ACA cohort. Collectively, the total costs for the group receiving CD as first-tier therapy were $74,329 lower than those receiving ACA ($218,756 vs. $293,084).

Univariate sensitivity analysis

Whether the efficacy of first-tier CD was decreased to that of ACA or the efficacy of ACA was increased to that of CD, CD resulted in lower costs and equivalent outcomes, suggesting that CD is consistently cost saving compared with ACA. In this particular instance, the cost savings are attributable to lower side-effect costs in the CD cohort.

Modifying the costs of the first- and second-tier therapies had a substantial impact on incremental costs, although not as much of an impact as modifying efficacy (Figure 2). Increasing or decreasing the cost of ACA by 15% resulted in differences of more than ± $11,000 from the base case of $74,328. Modifying the cost of CD had a slightly smaller effect; increasing or decreasing the cost of CD by 15% resulted in a window of more than ± $8,000 from the base case. The only other sensitivity analysis that had an effect of more than ± $5,000 was modifying the rate of side effects for ACA by ± 50%. Modifying other parameters, including the cost of General Practitioner (GP) and specialist visits by ± 15%, the rate of CD side effects by ± 50% and the cost of side effects (substituting the lowest and highest possible costs as reported in the physician survey), resulted in small effects on the results.

Figure 2. Univariate sensitivity analysis Tornado diagram.

Probabilistic sensitivity analysis

For the probabilistic sensitivity analysis (PSA), statistical distributions were substituted in the deterministic model to represent uncertainty in the following model parameters: CD and ACA AOMT cure rates; drug, physician and laboratory costs; and cost associated with the treatment of drug-related side effects. Drug costs were varied using a triangular distribution where the mean values of the deterministic model were used as the likeliest values and the 2006 Red Book price extremes were used as the minimum and maximum values. CD and ACA cure probabilities were allowed to vary within a triangular distribution encompassing the 95% confidence limits (CL) of the mean values of the deterministic model. Physician costs, laboratory costs and cost associated with drug side effects were varied within a normal distribution with a mean value equal to the deterministic model value and a standard deviation equal to 20% of their corresponding mean values.

One million second-order Monte Carlo simulations were performed to calculate the expected mean costs of CD and ACA therapy and their corresponding mean 95% CL presented as estimates per cohort of 1,000. In addition, standard deviations as well as minimum and maximum costs per cohort of 1,000 are provided. Lastly, the mean incremental cost of ACA therapy compared with CD therapy is also provided in the PSA results. Table 4 presents the results of the PSA.

Table 4. CD–ACA probabilistic sensitivity results*.

	Mean	sd	Lower mean 95% CL	Upper mean 95% CL	Minimum	Maximum
CD	218,832	752	218,785	218,878	216,058	221,236
ACA	293,186	821	293,136	293,237	290,360	295,827
ACA–CD	74,355	657	74,314	74,395	72,282	76,875

CD, ciprofloxacin/dexamethasone; ACA, amoxicillin/clavulanic acid; sd, standard deviation; CL, confidence limit.

* All costs are presented in US$ and expressed as costs per cohort of 1,000.

The PSA indicated that CD therapy provides cost savings compared with ACA therapy 100% of the time ( Table 4

 Google Scholar

) given the modelled parameters. The lack of overlap between the CD and ACA mean 95% CL is a clear indicator of the economic dominance of ototopical CD over oral ACA with respect to AOMT therapy. The cost savings associated with CD therapy was estimated at $74,355 per cohort of 1,000. Figure 3 illustrates the incremental cost-savings distribution associated with CD therapy compared with ACA therapy based on the Monte Carlo simulation. The CD cost-savings distribution (Figure 3) represents the range and probability of economic gain associated with initiating AOMT therapy with ototopical CD as opposed to initiating therapy with oral ACA. The probabilistic sensitivity analysis supports the deterministic model in the assertion that ototopical CD is more cost effective than oral ACA for the initial treatment of AOMT.

Figure 3. CD cost-savings distribution histogram.

Discussion

CD ear drops were shown to be cost effective compared with oral ACA in children with AOMT. The choice of first-tier therapy had an important impact on the overall success and cost per success. Whilst no therapy boasts perfect clinical success, initiating therapy with a more effective agent can reduce the costs associated with follow-up lines of therapy, including additional physician visits, testing and resources used to treat side effects, thus lowering treatment costs over a cohort of patients.

The cost-effectiveness advantage of CD can be attributed to two factors: (1) effectiveness; and (2) the side-effect profiles of ototopical versus oral antibiotic therapy. A difference in cure of 26.1% in the first-tier of therapy resulted in substantially lower downstream re-treatment costs in the CD pathway. Conversely, the high prevalence of side effects associated with ACA add considerable economic burden to the ACA pathway. When examined collectively, the effectiveness and side-effect profiles of the initial therapies initiate a cascade of cost consequences that favour CD and persist across the entire clinical algorithm despite the higher acquisition cost of CD. These findings were consistent across an extensive range of parameters, as the univariate and PSA demonstrated.

This model has several limitations. First, much of the data were gathered from a physician survey. Whilst this survey was large (300 participants), diverse (included GPs, paediatricians and otolaryngologists) and there were few skipped items (it was presented online and responses were required for most items), recall of clinical success rates or treatment of side effects may not be entirely accurate. Bias is expected to be similar across treatments. However, the use of a survey is also a strength: by using a survey-based approach rather than guidelines, the model captures not best practices but rather real-world practices. Furthermore, the survey is a unique source of data on real-world patterns of treatment for side effects. The impact of potentially atypical treatment patterns in the physicians’ responses is likely to affect parameters for both model choices equally.

The model also only follows patients through third-line therapy, although data suggest that there may be additional treatments required for a small number (<1%) of patients. Based on the physician survey, the rates of hospitalisation were negligible and thus hospitalisation was not included in the model. Thus, by not including hospitalisation, the model may underestimate the cost differential between treatments.

The higher initial cure rate with CD may result in fewer missed work days for the parent/caregiver. As missed work can have a substantial impact on overall costs of illness16, the fewer lines of therapy needed to cure, suggesting fewer visits to the doctor, is also an important economic advantage for CD. As data were not available, this model was unable to estimate the indirect costs associated with lost work and/or daycare due to repeated visits, thus the model may underestimate the total costs associated with treatment.

The model illustrates that therapy with CD is cost saving for AOMT compared with ACA. The modelled cost savings are insensitive to realistic changes in clinical cure rates and costs. Consequently, the results indicate a robust model that consistently finds CD to be cost saving compared with ACA for treatment of AOMT in children.

Acknowledgements

Declaration of interest: Jordana K Schmier is an employee of Exponent and received a grant from Alcon Research to develop this model and manuscript. Curtis R Waycaster is a Senior Health Economist with Alcon Labs Inc., the manufacturer of Ciprodex (ciprofloxacin 0.3% and dexamethasone 0.1%). G Michael Wall is Senior Director of Pharmaceutical Development at Alcon Research and led the team that brought Ciprodex® Otic to market.

The authors appreciate input on the preparation of the physician survey from Peter Roland MD, Chair of Otolaryngology, University of Texas Southwestern Medical School, Dallas, TX, USA.

Notes