| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Pharmacoepidemiology Unit, EA643 and Poison Centre, 2 Clinical Pharmacology Unit, EA643, ClaudeBernard University and 3 Respiratory Medicine, South Lyon Hospital Centre University Hospital, Lyon, France. 4 Merck & Co., Inc., Whitehouse Station, NJ, USA. 5 Pharmacoepidemiology Unit, McGill University, Montreal, Canada
CORRESPONDENCE: E. Van Ganse, Centre Antipoison, Centre de Pharmacovigilance, Pavillon N, Hôpital Edouard Herriot, Lyon, 69347, Cedex 03, France. Fax: 33 472666444. E-mail: eric.vanganse@chulyon.fr
Keywords: asthma, control, health economics, resource utilisation, severity, therapy
Received: December 12, 2001
Accepted April 4, 2002
The Pharmacoepidemiology Unit is part of the EA643 (ClaudeBernard University, Lyon, France) and of the Poison Centre (E. Herriot Hospital, Lyon, France). The study was supported by the Biomed grant Nr BMH4 CT965033 and by Merck & Co., Inc., Whitehouse Station, NJ, USA.
 |
Abstract |
|---|
 TOP Abstract Materials and methodsResultsDiscussionReferences |
|---|
2) in France. In 1998 a computerised family practice database was used to identify asthma patients aged 17–50 yrs. Information from the database was complemented by a patient survey to retrospectively assess the level of asthma control and hospital contacts. Costs of MRU over a 12month study period were related to demographics, medical history, asthma control, and doses of inhaled corticosteroids prescribed during the prestudy period.
A review of the computerised medical database identified 1,038 adult patients with persistent asthma, who completed the survey questionnaire. Over a 12month period, the mean cost of MRU was 549.8 E for wellcontrolled patients, 746.3 E per patient with moderate control, and 1,451.3 E per patient with poor control. Costs also increased significantly with age, access to free asthma care, comorbid conditions, asthma symptoms in the past year and whether inhaled corticosteroids had been prescribed before the study period.
In patients with persistent asthma, large differences were observed in the use of medical resources according to control and severity. Therefore, if patients appropriately use prescribed control therapy, their use of medical resources may be reduced.
Two factors should be distinguished in asthma: the degree of disease severity and the level of control achieved by the patient 1. Asthma severity can be considered as an intrinsic and rather stable characteristic of the disease for a given patient, i.e. the maximal level of impairment that asthma can bring to that patient, up to and including death; it is unclear how severity is affected by therapy 2. One way to assess severity is the level of corticosteroid prescription, inhaled or oral 3. By contrast, the level of control varies with time; it is primarily determined by contacts with allergens or infectious agents, and by therapy 4. In theory, the level of control can be influenced by patients, since they may influence two determinants of control, i.e. exposure to asthma triggers and adequate use of prescribed therapy. Control can also be influenced by how the healthcare system recognises and responds to patients who receive suboptimal therapy 5.
Many studies suggest that, on average, asthma patients do not use controller medicines optimally and that the resulting undertreatment puts the patient at risk of adverse outcomes, including asthma exacerbations, unscheduled visits to the consulting room, emergency room (ER) contacts and hospitalisations. A recent study identified additional risks factors for visits to the ER by asthma patients: absence of predetermined crisis plan, previous visits to ER facilities, and inadequate use of prophylactic medicines during a flareup of an exacerbation 6. Poor compliance with controller medicines has been identified as a major determinant of hospitalisation due to asthma 7. Drugutilisation studies have shown that medical resource utilisation (MRU) is increased threefold among high users of inhaled ß-agonists. Furthermore, among these patients, hospital costs appear to be determined by disease severity 8. In their study, LANES et al. 9 found that 8% of asthma patients had either a hospital admission or an ER visit during a 12month study period; their hospital costs accounted for 25% of the total costs of asthma care.
While studies confirm that a minority of patients use the majority of medical resources for asthma, the relationship between the use of medical resources and either control or severity needs clarification 10. Also, it is unclear whether the relationship between control and MRU varies according to the degree of asthma severity. The present study examined MRU in adult patients with persistent asthma in France. The objectives were to provide descriptive information about direct cost of asthma care, and to identify variables related to the costs of MRU, with a specific emphasis on the relationship between cost and level of control and its variation according to asthma severity.
|
Materials and methods |
|---|
|
TOPAbstractMaterials and methodsResultsDiscussionReferences |
|---|
Computerised database
BKLThalès, a computerised database of patients of French general practitioners (GPs) was accessed in early 1998. GPs who agreed to participate were located throughout the French territory. Patients with persistent asthma (Global Initiative for Asthma (GINA) stage 2) were identified in the database according to the following criteria: age 6–50 yrs, at least 18 months of followup by a BKLThalès GP, no prolonged intake of oral corticosteroids and use of at least three canisters of inhaled relievers within a 6-month period prior to the study period (baseline period). This paper presents data on adult patients (17 yrs).
Patient survey
Patients identified in the database received a survey form from their care giver. Patients answered questions on the level of asthma control (last 2 weeks), asthma symptoms during the last 12 months (occurrence of any asthma crisis? yes/no; occurrence of wheezing? yes/no), and occurrence of visits to ER and hospitalisation for asthma (last 12 months).
Study timelines
Details have been published previously 3. Briefly, patients were evaluated over at least 18 months, of which the 12 months preceding the survey were considered as the study period and prior months were considered as the baseline period. Figure 1
illustrates the study timelines.
|
Prescription of inhaled corticosteroids during the baseline period was used as an indicator of asthma severity: high (at least 800 µg of budesonide·day–1 or equivalent), moderate (<800 µg·day–1), low (no use) 3.
The method that was used to assess asthma control was inspired by the definition by COCKCROFT and SWYSTUN 1 i.e, minimal symptoms and minimal requirement of shortacting relievers: the algorithm for control used in this study is presented in table 1.
|
Specific costs were derived from MRU after multiplying each component by a unit cost for 1997, and were expressed in Euros (E). The unit cost estimate for visits to general practitioners (22.3 E) was obtained from the French Social Security (Caisse Nationale d'Assurance Maladie). The average cost for inpatient stay due to asthma in France (2,544.4 E) was obtained from the Programme de Médicalisation des Systèmes d'Information, the organisation responsible for computing diagnosisrelated group costs in France. This cost included all interventions applied during the hospital stay and inpatient therapy. The average cost for an asthmarelated ER visit (47.1 E) was determined by experts (University of Montpellier, Montpellier, France). Unit costs estimates for all prescribed medicines were obtained from the drug industry 12.
Data analysis
Data analysis was conducted in patients with information available from both BKLThalès and the survey. Analysis was conducted in two steps: first, a descriptive analysis of specific and total MRU costs and second, identification of correlates for total MRU cost. This analysis was focused on the relationship between total MRU cost and level of asthma control, allowing for an effect of asthma severity.
Descriptive information on costs
First, distributions of costs for each specific MRU (prescribed medicines, visits to GP, ER visits, hospitalisation) were studied for patients who completed the survey. These specific costs were conditional, i.e. each specific cost was computed only in patients who used the corresponding MRU during the study period. Next, total cost comprising the four MRUs was computed in patients who completed the questionnaire. When patients reported no visit to ER or no hospital admission for asthma, they were considered as having null expenses for this specific outcome. Similar assumptions were made for GP visits or prescriptions records in the database.
Correlates of total medical resource utilisation cost
Univariate analysis
Total MRU cost was studied separately across levels of control (good, moderate, poor), asthma severity (low, moderate, high) and other variables: sex, smoking (ever: yes/no), age (17–30, 31–40, >40 yrs), presence of comorbid conditions (yes/no), access to free asthma care (yes/no), asthma symptoms in past year (any crisis, any wheezing: yes/no). Wilcoxon and KruskalWallis tests were used to compare costs according to different classes of variables. The relationship between cost and level of control, stratified on the degrees of asthma severity was then studied.
Multivariate analysis
All multivariate analyses were carried out using multiple linear regression, with total MRU cost as the dependant variable. Total MRU cost was logtransformed to normalise its distribution, after discarding patients with null cost. Asthma control and asthma severity were each included in the model with three levels. All other variables associated with total MRU costs in univariate analyses at p<0.20 were also included in the model.
Two linear regression models were defined, M0 and M1. Each model contained asthma severity and asthma control and the cofactors mentioned above. M1 also included the asthma control by asthma severity interaction to assess whether the level of control was modified by the degree of severity, or vice versa. The F-test was used to test whether the inclusion of the interaction improved the goodness of fit. The M0 model was accepted as the final multivariable model if the interaction was not statistically significant.
|
Results |
|---|
|
TOPAbstractMaterials and methodsResultsDiscussionReferences |
|---|
.
|
Descriptive cost analysis
Specific conditional medical resource utilisation costs
The distributions of specific MRU costs are displayed in table 3.
|
). However, specific costs for these two resources accounted for only 17% and 33% of total MRU cost, respectively. By contrast, only 9% of patients were hospitalised but this resource made up 49% of global expenses (fig. 2). Visits to the ER had little impact on total MRU cost.
|
). A long distribution tail appeared beyond 1,200 E, and 97 patients had their costs concentrated beyond 2,000 E. Based on recent symptoms, these 97 patients had a significantly lower level of control compared to the others (p<0.0001); they were also more likely to have access to free asthma care and to report asthma crisis during the past 12 months (p<0.001). For these 97 patients, the part of MRU due to hospitalisation exceeded 86% of the total cost (376,578 E/434,309 E). In 1,038 patients, 29 had a null value for total cost since they had no asthmarelated MRU during the 12month study. These patients did not differ from their 1,009 counterparts with nonnull total MRU costs for any variable of the study except for comorbid conditions (p<0.0001): none of the nullcost patients had any comorbid diagnosis recorded.
|
.
|
Multivariate analysis
Multivariate analyses were conducted in 872 patients with nonnull costs (log transformation) and valid missing data for all variables included in models. Given the high correlation between the two variables reporting occurrence of asthma symptoms during the study period (Chisquared (one degree of freedom)=238, p<0.0001), "asthma crisis" was the only symptom variable included in the model, since it was more closely correlated to total MRU cost than "wheezing". Both M0 and M1 included age, presence of comorbid conditions, access to free asthma care, level of control and degree of severity. M1 included the four additional interaction terms. Accepting these four interaction terms did not yield significant improvement (F4,862=0.80, p=0.90); none of the four interaction coefficients reached significance (data not reported). Consequently, M0 was retained as the final multivariate linear model (table 5). Most univariate results were confirmed.
|
|
Discussion |
|---|
|
TOPAbstractMaterials and methodsResultsDiscussionReferences |
|---|
Both the level of control and degree of severity, as measured in this study, are correlates of MRU. Level of control and asthma severity were identified from distinct sources and at different points in time: prescription information for severity, i.e. the dose of inhaled corticosteroids prescribed by the family physician before the study period; and patient data for assessment of control, i.e. the patient's response to a specific questionnaire at the end of the study period. The level of control was assessed for a 2-week period at the end of the study period. However, this level of control may reflect the patients' attitude towards asthma medications drug observance and avoidance of risk factors throughout the study period. In a recent survey, VOLLMER et al. 13 found that a shortterm control index was significantly related to healthcare utilisation in a consecutive 12month period.
Asthma severity was assessed from average doses of prescribed inhaled corticosteroids during a baseline prestudy period (fig. 1). The marker of severity in this study was correlated with access to free healthcare for asthma (p=0.0004), number of prescriptions of oral corticosteroids (p=0.01) and number of different classes of asthma medications prescribed during the 12month study period (p<0.0001). Finally, the significant increase of MRU costs across degrees of severity is consistent with recent data from SERRABATLLES et al. 14. In that study, severity was based on the International Consensus Report on Diagnosis and Treatment for Asthma 15.
Control of asthma also had an important effect on costs, independently of severity. Poor control was associated with higher MRU in the preceding 12 months, suggesting that a significant part of the burden of asthma could be prevented with improved disease management. The comparison of MRU across levels of control within different degrees of severity is of interest. In all severity categories, costs increased significantly with decreasing levels of control. In each category, costs for "wellcontrolled" patients could be interpreted as indicative of the "minimal" costs of asthma care according to disease severity. A priori, such costs are nonpreventable, in contrast to the costs computed in groups with lesser levels of control.
Age correlation with cost is not surprising, since age is a known risk factor for resource consumption in asthma, as in other conditions 16. The significant relationship between comorbid conditions and cost was less expected in this rather young population (17–50 yrs) where asthma was the major morbidity. It suggests that conditions linked to asthma may have an impact on resource consumption 17. Concomitant psychiatric troubles have been shown to be associated with increased asthma morbidity, and they may contribute to increased costs 18. Having access to free asthma care also resulted in higher costs of MRU. These patients were older and had more comorbid conditions, but multivariate analyses suggested that access to free asthma care was associated with higher costs independent of these characteristics. This finding is in agreement with the purposes of access to free asthma care in France, provided to patients with more severe disease, independent of their socioeconomic status, after review of their medical history by Social Security experts.
Only crude comparisons can be made between the data presented here and information available for other countries. BARNES et al. 19 made a comparison of direct costs between countries; depending on the country and the study, hospital contacts comprised between 18–79% of direct costs, i.e. generally a lesser portion of the costs of MRU than in this study. BARNES et al. 19 found that improved control was associated with an increase in drug costs and more visits to the GP. It should be stressed that, as a result of the selection of patients with persistent asthma, the present data did not allow valid estimates of the total asthma burden in the French population, in contrast to data published by other investigators 19–21. Another effect of the selection of patients with persistent asthma may have been the relatively high incidence of hospital contacts. Nevertheless, this is also the result of efforts to collect data on this outcome directly from patients. The current authors have indeed demonstrated that care givers are not optimally informed of outcomes in asthma patients, and using family physician data may result in underestimates of the hospitalcontacts incidence 22.
These points underline some limitations of using a GP prescription database in asthma outcomes research. However, important advantages should also be acknowledged, i.e. that such databases may be considered as a more valid tool to assess asthma severity than drug dispensing databases, since data are available on actual GP's prescriptions of inhaled corticosteroids. Those prescriptions may be considered as being a reflection of the care givers' instant perception of asthma control, at least in settings and at times where such prescriptions were issued with some reluctance 23. A selection criteria prescribed was at least three canisters of inhaled ß-agonists during the 6-month period before the onset of the study. This could be considered as high ß-agonists consumption, but it was consistent with ß-agonists prescription patterns by GPs in France until recently.
This study had some other limitations. To compute medication costs, several assumptions were made, for example that patients took their prescriptions. Average costs were used to estimate the impact of admission to hospital for asthma, since there was no access to individual inpatient hospitalresource consumption data. Also, there may be recall bias in the responses to the survey questions, e.g. with some patients over or underestimating their rate of hospital admission. A selection bias could affect the results if there was a systematic, nonrandom, factor influencing the patient's decision to complete and return the questionnaire; 1,038 responders to questionnaires would then be different from 540 nonresponders. However, there were no significant differences between these two groups in demographic variables and asthma severity, though records of comorbid conditions were slightly higher among the 1,038 responders (p<0.04). Of 1,038 participants who completed the questionnaire, 101 did not answer specific questions for asthma control. This could represent another selection bias. However, patients with missing information on control did not differ from 937 others for costs or any variable in the study. If a selection bias cannot be excluded, its influence should be limited. Multivariate analyses were conducted in only 872 patients with nonnull total MRU cost and valid data for all covariates. Except for comorbid conditions, these 872 subjects did not differ from their 166 counterparts for any other studied variables (data not reported). Furthermore, all multivariate results were consistent with univariate analyses.
In the final linear model, global R2 did not exceed 0.17. Only a limited number of correlates for costs were identified. For example, the authors had no information on socioeconomic status, on education level, or on the attitude of the patients 24. However, a gradient of resource consumption was identified according to increasing severity or decreasing control, and this provided support to the assessment of these two diseases characteristics, and their relationship to resource consumption.
In summary, the results suggest that the major part of the cost burden of asthma is a consequence of both poor control and greater severity, especially for major outcomes, such as hospital admissions. Improvement in asthma control through improved disease management could potentially result in significant savings, in patient morbidity and costs to society 25. The findings also confirm that information obtained directly from patients is a useful addition to database information 3. The linkage of medical records to a specific patient survey may help to understand the complex relationship between severity and control in asthma, and to identify preferred areas of interventions to relieve the burden of asthma.
|
References |
|---|
|
TOPAbstractMaterials and methodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
%
%
%
Specific expenses in total medical resource utilisation cost 760,863E in 1,038 patients. : hospitalisation 49; : asthma medications 33; : visits to general practitioner 17; : visits to emergency room.