|
Strategies for diagnosing and treating suspected acute bacterial sinusitis: a cost-effectiveness analysis |
Balk E M, Zucker D R, Engels E A, Wong J B, Williams J W, Lau J |
|
|
Record Status This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn. Health technology Four diagnostic and treatment strategies for patients with suspected acute bacterial sinusitis (ABS) were compared. The strategies were:
no antibiotic treatment;
all patients given empirical antibiotic treatment;
clinical criteria-based antibiotic treatment; and
antibiotic treatment based on the results of radiography.
The antibiotic regimen was amoxicillin 250 mg, administered 3 times per day for 10 days.
Economic study type Cost-effectiveness analysis.
Study population The model used a hypothetical study population comprised of teenagers and adults with suspected, uncomplicated, community-acquired ABS, who were symptomatic for less than 4 weeks and who did not have recurrent sinusitis. Patients who were immunocompromised, or had atopy, malignancy, cystic fibrosis, or a history of sinus trauma or surgery, were excluded from the model.
Setting The setting was primary care. The economic evaluation was conducted in the USA.
Dates to which data relate The effectiveness evidence was derived from sources published between 1974 and 2000. The cost data were taken from sources published between 1997 and 2000. The prices used were not adjusted to a single price year.
Source of effectiveness data The effectiveness data were based on the non-systematic use of published studies and statistics.
Modelling A decision model was used to estimate the benefits and costs of the alternative patient management strategies for a time horizon of 14 days. A Markov process with a cycle length of 1 day was used, during which patients either remained in the same state of health, experienced serious complications from sinusitis, developed antibiotic side effects (if receiving antibiotics), or experienced an improvement in their symptoms. The model made the following assumptions:
there was no recurrence of resolved symptoms;
the risk of developing side effects from the antibiotics remained constant;
side effects lasted 2 days and required a change in prescription, but no further reactions occurred and the cure rate was unaltered;
antibiotic treatment of patients who did not have ABS did not alter their symptoms;
only untreated patients with ABS experienced severe disease complications (brain abscess, meningitis, facial or orbital cellulitis).
Outcomes assessed in the review The following parameters were obtained from an ad hoc review of the literature:
the prevalence of sinusitis;
the rate of antibiotic side effects;
the rate of sinusitis-related complications;
the sensitivity and specificity of applied clinical criteria;
the sensitivity and specificity of sinus radiography;
the cure rates on days 3, 7, 10 and 14 for the sub-groups of symptoms caused by ABS (with and without antibiotic treatment) and for symptoms caused by other diseases;
the number of symptom-free days;
the utility of mild, moderate and severe sinusitis for patients cured (with and without antibiotic side effects) and for patients sick (with and without antibiotic side effects); and
serious complications due to sinusitis.
Study designs and other criteria for inclusion in the review Although the review was ad hoc, the authors expressed a preference for using the results of recent meta-analyses, when available. Only two such studies were identified. The data from these studies were supplemented by individual studies and published specialist opinion.
Sources searched to identify primary studies Criteria used to ensure the validity of primary studies Methods used to judge relevance and validity, and for extracting data Number of primary studies included The review comprised 20 studies, including 2 meta-analyses and 1 published specialist opinion.
Methods of combining primary studies The methods used to combine the data were not reported. For some parameters only one source was cited. The range of results within the studies formed the range used in the sensitivity analyses of some of the parameters.
Investigation of differences between primary studies Results of the review The prevalence of sinusitis was 50% in the base-case and 0 to 100% in the sensitivity analysis.
The rate of antibiotic side effects was 5% in the base-case and 20% in the sensitivity analysis.
The rate of sinusitis-related complications was 1:10,000 in the base-case and 1:1,000 in the sensitivity analysis.
The sensitivity and specificity of applied clinical criteria were, respectively, 0.96 and 0.77 in the base-case and 0.60 and 1.00 in the sensitivity analysis.
The sensitivity and specificity of sinus radiography were, respectively, 0.90 and 0.61 in the base-case and 1.00 for both in the sensitivity analysis.
The cure rates for amoxicillin were 2% on day 3, 24% on day 7, 54% on day 10, and 87% on day 14.
The cure rates for amoxicillin resistance were 1% on day 3, 14% on day 7, 35% on day 10, and 68% on day 14. Similarly, the cure rates for symptomatic treatment were 0% on day 3, 5% on day 7, 15% on day 10, and 41% on day 14. When antibiotics were ineffective because symptoms were caused by other diseases, the cure rates were 35% on day 3, 61% on day 7, 75% on day 10, and 84% on day 14.
Overall, each day that a patient did not experience symptoms of either sinusitis or antibiotic side effects had a value of 1. Each day that a patient experienced symptoms of either sinusitis or antibiotic side effects, or suffered serious complications of sinusitis, had a value of 0.
The number of symptom-free days for patients cured without antibiotic side effects was 1.
Measure of benefits used in the economic analysis The measures of benefit used were the number of symptom-free days and the number of quality-adjusted symptom-free days. Utilities for each of the three levels of sinusitis severity were derived from the Quality of Well-being General Health Policy Model.
Direct costs The effectiveness and cost parameters entered in the model were reported separately. Only the direct costs of medical care were included in the analysis. The costs included were for antibiotics, radiography, antibiotic side effects, persistent sinusitis and serious disease complication. The cost of antibiotic side effects covered the return medical consultation, treatment of side effects, and change in medication to folate inhibitor or more expensive antibiotic). The cost of persistent sinusitis covered the return medical consultation and amoxicillin, folate inhibitor, or more expensive antibiotics. The costs of a serious disease complication comprised hospital costs, intravenous antibiotics and surgery. The direct cost data were obtained from three sources published between 1997 and 1998. These were not adjusted to a specific price year. Discounting was not applied, which was correct since the time horizon was 14 days. The median costs were reported. The authors did not include the costs of an initial medical consultation or symptomatic therapies, as these were assumed to be common to all management strategies.
Statistical analysis of costs Point estimates were presented. No statistical analysis of the costs was reported.
Indirect Costs The authors justified their reporting of the indirect costs on the grounds that the high frequency of sinusitis type symptoms resulted in high costs in work time lost. The lost earnings for a day of work were reported. Weekend days were valued at the same rate to reflect non-work productivity losses and other, unspecified, indirect costs. Productivity losses were calculated from data on the usual weekly earnings for wage and salary earners, obtained electronically from the Bureau of Labor Statistics. The number of days of work lost for each severity of sinusitis was derived using the model. A median value of $115 per day was used. Only productivity losses of patients, not their caregivers, were included. Discounting was not applied, which was correct since the time horizon was 14 days. Values for lost earnings per day were for the first quarter of 2000.
Sensitivity analysis Sensitivity analyses were undertaken to investigate the generalisability of the results. The parameters explored were event probabilities, diagnostic test performance, bacterial resistance and costs. In general, the authors did not describe how the ranges over which the variables were tested were chosen. However, the range for bacterial resistance was based on the lower 95% confidence intervals of a Kaplan-Meier curve and the effectiveness of newer antibiotics came from a published source. Although the authors did not specify the type of sensitivity analyses undertaken, it would appear that parameters were varied individually in one-way sensitivity analyses.
Estimated benefits used in the economic analysis The number of symptom-free days and the number of quality-adjusted symptom-free days were reported for the four management strategies at a disease prevalence rate of 50%.
Overall, treating all patients with antibiotics was associated with the most number of symptom-free days (6.6 days), followed by clinical criteria-guided treatment (6.5 days), radiography-guided antibiotic treatment (6.4 days) and no antibiotic treatment (5.0 days).
For mild symptoms, radiography-guided antibiotic treatment, treating all patients with antibiotics and treatment guided by clinical criteria all resulted in 12.4 quality-adjusted symptom-free days, compared with 12.2 days for the no antibiotic treatment strategy.
For moderate symptoms, radiography-guided antibiotic treatment, treating all patients with antibiotics and treatment guided by clinical criteria all resulted in 10.8 quality-adjusted symptom-free days, compared with 10.2 days for the no antibiotic treatment strategy.
For severe symptoms, radiography-guided antibiotic treatment, treating all patients with antibiotics and treatment guided by clinical criteria all resulted in 10.3 quality-adjusted symptom-free days, compared with 9.7 days for the no antibiotic treatment strategy.
The side effects of treatment were considered in the economic analysis.
Cost results The average cost of each of the four management strategies was reported.
Overall, and for severe symptoms, the treatment costs were $883 using clinical criteria, $886 for treating all patients with antibiotics, $974 for radiography-guided antibiotic treatment and $1,053 for no antibiotic treatment.
For mild symptoms, clinical criteria-guided treatment cost $26, empirical antibiotic treatment for all patients cost $38, radiography-guided antibiotic treatment cost $107 and no antibiotic treatment cost $20.
For moderate symptoms, clinical criteria-guided treatment cost $454, empirical antibiotic treatment for all patients cost $462, radiography-guided antibiotic treatment cost $541 and no antibiotic treatment cost $536.
Synthesis of costs and benefits The costs and benefits were combined by calculating the marginal cost-effectiveness of each management strategy, overall and at differing levels of symptom severity, assuming a disease prevalence rate of 50%.
The authors calculated that, overall, clinical criteria-guided treatment cost the least. However, treating all patients with antibiotics was slightly more effective than clinical-criteria guided treatment at a cost of $74 per symptom-free day. Both radiography-guided antibiotic treatment and no antibiotic treatment cost more and were less effective.
For mild symptoms, no antibiotic treatment cost the least, but clinical criteria-guided treatment improved effectiveness at a marginal cost-effectiveness of $34 per symptom-free day. The marginal cost-effectiveness of treating all patients with antibiotics rose to $22,800 and sinus radiography was more costly and less effective than the other strategies.
For moderate and severe symptoms, clinical criteria-guided treatment cost the least, but treating all patients with antibiotics was more costly and more effective. The marginal cost-effectiveness ratios ranged from $152 to $421 per symptom-free day. Both radiography-guided antibiotic treatment and no antibiotic treatment cost more and were less effective.
The results were sensitive to variations five parameters. More specifically, the severity of sinusitis symptoms, increased bacterial resistance, the use of newer and more expensive antibiotics, clinical criteria test performance and sinus radiography test performance.
The more severe the sinusitis symptom experienced, the greater the benefit of antibiotic treatment. The optimal treatment and diagnosis strategy altered with disease prevalence. As prevalence increased, strategies resulting in antibiotic treatment increased in overall benefit. The effect of increased bacterial resistance was to decrease the efficacy of the antibiotic in comparison with placebo, making all strategies more costly and less effective. Newer, more expensive antibiotics were not cost-effective to administer to all patients unless prevalence exceeded 80%. Clinical criteria remained optimal for sensitivities above 93% and specificities above 72%. Reducing the sensitivity to 95% led to empirical antibiotic treatment becoming cost-effective at lower prevalence levels. Sinus radiography was not cost-effective, even with 100% sensitivity and specificity, unless the total cost was less than $19.
Authors' conclusions Using a threshold of $50,000 per quality-adjusted life-year, ($137 per quality-adjusted day), the authors drew the following conclusions about the cost-effectiveness of each strategy. The optimal strategy depended on disease prevalence. Antibiotic treatment based on clinical criteria was cost-effective for mild and moderate symptoms where the prevalence of sinusitis was between 15 and 93% (mild symptoms) or between 3 and 63% (moderate symptoms), respectively. The use of clinical criteria-based treatment was cost-effective for severe symptoms at a disease prevalence of 51%, and for the prevention of sinusitis or antibiotic side effects at a disease prevalence of 44%. Sinus radiography-guided treatment was never cost-effective for initial treatment. In general, no antibiotic treatment was the preferred strategy when prevalence was low, and treating all patients with antibiotics was optimal when prevalence was high.
CRD COMMENTARY - Selection of comparators A justification was given for the comparators used. They reflected the current medical uncertainty about the best strategy for the management of patients with symptoms of ABS. You should decide if these are representative of diagnosis and treatment strategies for this disease in your own setting.
Validity of estimate of measure of effectiveness A systematic review was not undertaken. Although this is common practice with models, it does not always ensure that the best data available are used in the model. The authors appear to have used the data from the available studies selectively. They did not consider the impact of differences between the identified studies when estimating effectiveness.
Validity of estimate of measure of benefit The measures of benefit used were the symptom-free days and quality-adjusted symptom-free days, over 14 days. These were derived from the Quality of Well-being General Health Policy Model. The estimation of benefits was modelled. The instrument used to derive a measure of health benefit, a decision analysis model incorporating a Markov process, was appropriate.
Validity of estimate of costs The analysis of costs was performed from a societal perspective. It appears that all the relevant categories of costs have been included in this analysis. Some relevant costs were omitted from the analysis. The authors did not include the costs of an initial medical consultation or symptomatic therapies, as these were assumed to be common to all management strategies. The direct non-medical costs to the patients were also excluded. Although some costs were omitted from the analysis, their omission is unlikely to have affected the authors' conclusions. The costs were reported separately from other model parameters, thus enhancing the reproducibility of the study in other settings.
A sensitivity analysis was conducted on many of the model input parameters. The ranges used appear to have been appropriate. A Weibull function was fitted to Kaplan- Meier curves to estimate the daily rate of cure.
The costs were treated deterministically, but sensitivity analyses were conducted to assess the robustness of the estimates used. Discounting was not applied, which was appropriate given the 14-day time horizon. Costs, rather than charges, were reported. The cost data were taken from sources published between 1997 and 2000. However, no adjustments to a single price year appear to have been made.
Other issues The authors did not make appropriate comparisons of their findings with those from other studies. They reported that no randomised clinical trial that has taken all variables into consideration has been performed. The authors did not directly address the issue of the generalisability of the results to other settings. However, they explored the effect of disease prevalence in detail. The authors do not appear to have presented their results selectively. The study included adults and teenagers. Hence, this might have resulted in the overestimation of productivity losses since the majority of teenagers would not be in full-time paid employment. The costs were driven largely by symptom duration because of the high cost of productivity losses in comparison with the costs of treatment and diagnosis. The authors discussed the long-term effects of the loss of antibiotic efficacy due to increasing bacterial resistance. They anticipated that this would substantially increase costs while decreasing the benefits of using the clinical criteria and empirical antibiotic strategies.
Implications of the study The authors stated that an accurate, low-cost diagnostic test for ABS is needed. They suggested that further research is needed to investigate the prevalence of ABS in various settings, the efficacy of specific symptomatic treatments, and the rate of disease complications. Other suggestions were for further research on the effect on cure rate of disease complications, the effect on cure rate of increasing bacterial resistance, and the effect on clinical methods of diagnosing disease.
Source of funding This article is based on an evidence report produced by the New England Medical Center's Evidence-based Practice Center under AHRQ contract #290-97-0019. Additional support was provided by an NRSA training grant (#T32 HS00060) and AHRQ grant #R25 HS09796.
Bibliographic details Balk E M, Zucker D R, Engels E A, Wong J B, Williams J W, Lau J. Strategies for diagnosing and treating suspected acute bacterial sinusitis: a cost-effectiveness analysis. Journal of General Internal Medicine 2001; 16(10): 701-711 Indexing Status Subject indexing assigned by NLM MeSH Acute Disease; Anti-Bacterial Agents /therapeutic use; Cost-Benefit Analysis; Decision Support Techniques; Humans; Markov Chains; Quality-Adjusted Life Years; Sinusitis /diagnosis /drug therapy /economics /radiography; Treatment Outcome AccessionNumber 22001001997 Date bibliographic record published 31/05/2005 Date abstract record published 31/05/2005 |
|
|
|