The circles indicate the base-case risk estimate. This is made up of various means taken from the literature. The limit bars represent the risk values when the extreme maximum and minimum values of each input are used. E indicates embolization; P, posterior packing; and T, transnasal endoscopic sphenopalatine artery ligation.
This first-order sensitivity index shows the variances in the total risk is most sensitive to variances in cerebrovascular (CV) event risk. MI indicates myocardial infarction.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Leung RM, Smith TL, Rudmik L. Developing a Laddered Algorithm for the Management of Intractable Epistaxis: A Risk Analysis. JAMA Otolaryngol Head Neck Surg. 2015;141(5):405–409. doi:10.1001/jamaoto.2015.106
For patients with epistaxis in whom initial interventions, such as anterior packing and cauterization, had failed, options including prolonged posterior packing, transnasal endoscopic sphenopalatine artery ligation (TESPAL), and embolization are available. However, it is unclear which interventions should be attempted and in which order. While cost-effectiveness analyses have suggested that TESPAL is the most responsible use of health care resources, physicians must also consider patient risk to maintain a patient-centered decision-making process.
To quantify the risk associated with the management of intractable epistaxis.
Design and Setting
A risk analysis was performed using literature-reported probabilities of treatment failure and adverse event likelihoods in an emergency department and otolaryngology hospital admissions setting. The literature search included articles from 1980 to May 2014. The analysis was modeled for a 50-year-old man with no other medical comorbidities. Severities of complications were modeled based on Environmental Protection Agency recommendations, and health state utilities were monetized based on a willingness to pay $22 500 per quality-adjusted life-year. Six management strategies were developed using posterior packing, TESPAL, and embolization in various sequences (P, T, and E, respectively).
Main Outcomes and Measures
Total risk associated with each algorithm quantified in US dollars.
Algorithms involving posterior packing and TESPAL as first-line interventions were found to be similarly low risk. The lowest-risk approaches were P-T-E ($2437.99 [range, $1482.83-$6976.40]), T-P-E ($2840.65 [range, $1136.89-$8604.97]), and T-E-P ($2867.82 [range, $1141.05-$9833.96]). Embolization as a first-line treatment raised the total risk significantly owing to the risk of cerebrovascular events (E-T-P, $11 945.42 [range, $3911.43-$31 847.00]; and E-P-T, $11 945.71 [range, $3919.91-$31 767.66]).
Conclusions and Relevance
Laddered approaches using TESPAL and posterior packing appear to provide the lowest risk. Combining risk and cost-effectiveness perspectives, we recommend a laddered approach to intractable epistaxis with TESPAL first, followed by either embolization or posterior packing.
Intractable epistaxis presents as one of the most common otolaryngologic emergencies. When anterior nasal packing fails, posterior nasal packing is used to stabilize patients.1 Historically, management has involved prolonged posterior packing for 2 to 7 days, with transnasal endoscopic sphenopalatine artery ligation (TESPAL) or endovascular embolization reserved as second- and third-line therapies. However, posterior nasal packing causes considerable discomfort, requires prolonged hospitalization, and has bleeding recurrence rates as high as 52%.2 Meanwhile, TESPAL and embolization rebleeding rates have been found to be less than 10%.3,4
Recently, an economic evaluation by Dedhia et al5 showed that it was more cost-effective to proceed directly to TESPAL once patients are stabilized with a posterior pack, compared with prolonged posterior packing. Meanwhile, a recent modeling-based economic evaluation by Rudmik and Leung6 have demonstrated that TESPAL was more cost-effective than embolization as a primary treatment for intractable epistaxis. While it is important to identify the most cost-effective management strategies for a sustainable, fiscally responsible medical system, it is critical for physicians to maintain a patient-centered focus. As physicians, we must provide patients with effective treatment while simultaneously minimizing exposure to adverse events. This becomes an issue of risk.
The objective of this study was to evaluate the risk profile of 6 laddered management algorithms involving posterior packing (×2-7 days) (P), TESPAL (T), and embolization (E) as first-, second-, and third-line interventions in different sequences: (1) P-T-E, (2) T-E-P, (3) E-T-P, (4) P-E-T, (5) T-P-E, and (6) E-P-T (second and/or third procedures are performed if the previous procedure proves to be unsuccessful). We hypothesize that a ladder involving TESPAL as a first-line approach will be the preferred management strategy to minimize patient risk.
This modeling-based simulation was performed for a 50-year-old patient with refractory epistaxis despite anterior nasal packing, who had no relevant medical comorbidities and was not taking any anticoagulant medications. The analysis took the perspective of the patient for this analysis. Excel 2010 (Microsoft Corp) was used to perform the calculations in the simulations.
A risk model was constructed using the probabilities of the severe and/or permanent adverse events associated with each intervention. A literature search was performed using PubMed and MEDLINE using the following terms: epistaxis, posterior packing, sphenopalatine artery ligation, and embolization. The literature search included articles from 1980 to May 2014. A total of 422 results were returned and reviewed. Probabilities were extracted from the literature and are summarized in Table 1. Adverse events were assembled from the literature and assigned probabilities and impacts based on Environmental Protection Agency (EPA) recommendations (Table 2).11-14 Adverse events used to build the model are given in Table 1. The search focused on systematic reviews and randomized clinical trials whenever possible. The reference lists of all studies were examined to ensure all relevant studies were captured.
To quantify risk and allow for a common comparison of different health states, the impact of each complication was monetized into US dollar amounts. This method also allows for impacts such as lost income, along with immediate and ongoing costs of medical care, to be accounted into the quantification of risk. This is a commonly used modeling technique in disaster relief planning.11-14
The severity of each adverse effect was monetized based on the EPA recommendations of a single death valued at $10 million11-14 and varying levels of morbidity at appropriate fractions of this overall cost (Table 2). For example, the risk of cerebrovascular event would represent a severe irreversible disability valued at $1 million based on the EPA recommendations. To corroborate these estimates, health state utility values for each adverse event were converted into US dollars based on the willingness-to-pay literature of $22 500 per quality-adjusted life-year (QALY) (range, $12 500-$32 500).21 Ahlsiö et al18 identified a Health Utility Index 3 loss of 0.26 for a cerebrovascular event. Therefore, for a 50-year-old individual to reach the average life-expectancy age of 80 years, this would produce 7.5 QALYs lost, or $175 500. Adding in the median US annual income of $52 000 per year ($20 500 for the 20% percentile, $104 000 for the 80% percentile)22 lost income from age 50 years to retirement at age 67 years would produce another $884 000. This would bring the total to approximately $1 million. Lost productivity was modeled at $1000 per week, corresponding to 2012 median US income levels.22
Risk is calculated as a product of the probability of an adverse event occurring multiplied by the impact of that event. The total risk can be expressed as a sum of the individual adverse event risks. The proportion of patients in whom initial interventions had failed undergo further intervention and the associated risk. Examples of algorithms can be expressed as formulae, as given in the Box.
pPP = Probability of adverse event occurring after embolization
iPP = Impact of posterior packing adverse events
fPP = Failure rate after posterior packing
pEmbol = Probability of adverse events occurring after embolization
iEmbol = Impact of embolization adverse events
fEmbol = Failure rate after embolization
pTESPAL = Probability of adverse events occurring after TESPAL
iTESPAL = Impact of TESPAL adverse events
FTESPAL = Failure rate after TESPAL, major bleeding requiring embolization
fTESPAL = Failure rate after TESPAL, minor bleeding requiring packing
Due to the inherent uncertainty of the estimates, a sensitivity analysis was performed to identify an upper and lower range of risk for each algorithm. When the literature search identified a single source for any adverse event, the range was calculated with 1 more or 1 less event occurring in that population.
A first-order sensitivity index was also calculated and compiled to identify the significance of individual adverse events.13 This is a subanalysis used to characterize how variances in individual adverse events affect the total variance in each model. This is useful to identify which adverse events are most likely to contribute to deviations in the model. It helps identify which uncertainties are most responsible for the uncertainties in the model. This is calculated by identifying the change in risk when an individual factor is set to its maximum. This change is then expressed as a percentage of the total change in risk when all factors are set to their maximums.
From an effectiveness perspective, all 6 laddered strategies would result in greater than 99% resolution of epistaxis after 2 interventions. However, the laddered strategies starting with TESPAL and embolization are more likely to succeed with a single procedure. When evaluating overall risk of each laddered strategy, the P-T-E, T-P-E, and T-E-P strategies were similar in risk profiles—all had risks lower than $4100 (Figure 1). The P-E-T laddered strategy had an intermediate risk profile (mean risk of $6786), while the 2 highest-risk strategies were those using embolization as a first-line intervention (E-T-P and E-P-T), with both having risks greater than $12 000 (Figure 1).
A first-order sensitivity index was compiled to identify major contributors to variance. A graphical example of this analysis is shown for E-T-P (Figure 2). The effects of individual risk factors were expressed as a fraction of the maximum range estimate for each strategy when individual risk factors were maximized. P-T-E was sensitive to the uncertainties in pain/paresthesia after embolization (Figure 2). Meanwhile, algorithms with a high likelihood of progressing to embolization (P-E-T, E-T-P, and E-P-T) were highly sensitive to cerebrovascular complications.
While it is important to maintain socially responsible practice by implementing cost-effective interventions, physicians must also provide patient-centered care. This study developed a model to assess the patient risk associated with 6 different laddered approaches to the management of intractable epistaxis. The outcomes suggest that the P-T-E, T-P-E, and T-E-P are low-risk management strategies compared with using strategies that use embolization as a first-line intervention. This is primarily due to a relatively high stroke risk associated with embolization. The top 3 algorithms are separated by a difference of approximately $1000 in risk, which is equivalent to 2 weeks of life in the willingness-to-pay literature—a clinically small difference. Effectively, the P-T-E, T-P-E, and T-E-P algorithms carry equivalently low-risk profiles.
Results from this modeling-based risk assessment study can combined with evidence on cost-effectiveness to help clinicians in the decision-making process for management of intractable epistaxis. To reconcile the risk and cost-effectiveness perspectives, we suggest a management ladder starting with TESPAL as a first-line treatment. Patients who continue to bleed should proceed to either posterior packing or embolization as a second-line treatment, depending on the availability of local resources.
This study is limited by its ability to estimate the true probabilities and impacts of the various complications. Wide ranges of estimates have been incorporated into the model to account for this potential uncertainty in risk conclusions. Even under the extremes of the sensitivity analysis, the embolization-first strategies (E-P-T and E-T-P) are the highest-risk options, while both TESPAL-first strategies (T-E-P and T-P-E) and 1 posterior packing–first strategy (P-T-E) have the lowest-risk profiles. The first-order sensitivity index revealed that the dominant risk factor against embolization was the risk of cerebrovascular injury. Embolization-first strategies remain the highest-risk algorithms unless the cerebrovascular risk decreases by a factor of 10.
A further limitation of this study lies with the differences in patient-specific tolerance for each complication. Patients of different ages, occupations, and cultural backgrounds may have different tolerances for each adverse event. Although this study was conducted with this in mind, it is a tool for clinicians to aid in decision making. It should not replace a comprehensive discussion of treatment options, risks, and benefits. Lastly, it is important to remember that this model assumed that patients did not have any significant medical comorbidities (ie, uncontrolled hypertension, cardiac disease, or severe mid-face trauma) and were not receiving anticoagulant therapy. The quantified risks reported in this study may not be appropriate in the presence of any of these factors and may lead the clinician to make different management decisions.
To provide a patient-centered perspective for the management of intractable epistaxis, this study has performed a risk analysis of 6 laddered treatment strategies. On the basis of the results from this modeling-based analysis, the lowest-risk laddered strategies appear to be those that use TESPAL as a first-line option or posterior packing following by TESPAL. When the risk outcomes are combined with cost-effectiveness data, we believe that TESPAL should be considered the first-line strategy to minimize patient risk and promote the most efficient use of health care resources. It should be recognized that the published studies are the results of experienced surgeons. Not all institutions may possess expertise in each modality of treatment; it may be reasonable to refer to tertiary centers as required.
Submitted for Publication: December 5, 2014; final revision received December 29, 2014; accepted January 12, 2015.
Corresponding Author: Randy M. Leung, MD, Department of Otolaryngology–Head & Neck Surgery, University of Toronto, Royal Victoria Regional Health Centre, 125 Bell Farm Rd, Ste 203, Barrie, ON L4M 6L2, Canada (email@example.com).
Published Online: February 26, 2015. doi:10.1001/jamaoto.2015.106.
Author Contribution: Dr Leung had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: Leung, Rudmik.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Leung, Rudmik.
Administrative, technical, or material support: Leung.
Study supervision: Smith.
Conflict of Interest Disclosures: None reported.
Create a personal account or sign in to: