Outcomes and cost benefits associated with the introduction of inhaled corticosteroid therapy in a medicaid population of asthmatic patients

CLINICAL THERAPEUTICSWOL.

20, NO. 3, 1998

Outcomes and Cost Benefits Associated with the Introduction of Inhaled Corticosteroid Therapy in a Medicaid Population of Asthmatic Patients Rajesh Balkrishnan, MS (Pharm),’ G. Joseph Notwood, PhD,’ and Andrea Anderson, Pharmp ‘Division of Pharmaceutical Policy and Evaluative Sciences, School of Pharmacy, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, and 2Clinical Applications Research, US Medical Affairs, Glaxo Wellcome Research and Development, Research Triangle Park, North Carolina

ABSTRACT A retrospective cohort study was conducted to assess the clinical and economic impact of the introduction of inhaled corticosteroid therapy in the North Carolina Medicaid population of patients with asthma. The case group consisted of 180 patients who were followed for 1 year before and 1 year after the initiation of inhaled corticosteroid therapy. The control group consisted of 233 patients whose asthma was of similar severity to that of the case group and who remained on any therapy other than corticosteroids for a continuous 2-year period. After the initiation of inhaled corticosteroid therapy, the case group had reductions of 50% in hospitalizations, 26% in outpatient visits, and 15% in physician visits. At the end of the second year, the control group had significant increases of 23% in hospitalizations and 36% in outpatient visits. All of the changes were statistically significant.

0149-2918/98/$19.00

However, after adjusting for potential confounding factors, there was a nearly 24% decrease in total health care costs per asthmatic patient per month as a result of the introduction of inhaled corticosteroid therapy. Overall, we found that there was a cost benefit to Medicaid in the introduction of inhaled corticosteroid therapy and that this therapy brought about dramatic decreases in health care utilization and costs. Key words: outcomes, cost benefit, inhaled corticosteroids, Medicaid.

INTRODUCTION Asthma is a chronic condition characterized by excessive airway narrowing in response to a wide range of commonly occurring stimuli. It is relatively common in the United States, estimated to affect between 9 and 12 million persons (about 4% of the population).’ Not only is asthma one of the most common chronic diseases of children and adults. but there have been

567

CLINICAL THERAPEUTICS”

large increases in its prevalence and severity in the past few years2 Asthma is associated with significant morbidity and mortality, increasing rates of which have been attributed to ineffective management,3 including underutilization of corticosteroids and improper follow-up of therapy.2 Drug therapy is an important factor in the management of asthma, and appropriate drug therapy can significantly improve therapeutic outcomes and quality of life. On the other hand, less-than-optimal drug therapy can have significant medical consequencesincluding treatment failure and drugrelated morbidity and mortality-that can adversely affect patients’ economic, social, and psychological lives. In addition to direct morbidity from taking drugs inappropriately, noncompliance can contribute to as many as 25% of all hospital admissions and treatment failures.4 One way of optimizing the management of asthma is with inhaled therapy. The use of bronchodilators and corticosteroids in the form of inhaled therapy is one of the most prevalent and preferred modes of treatment. This dosing form offers the advantages of rapid onset of action compared with orally administered drugs, delivery of drug directly to the lung, low incidence of side effects, small doses, and convenience of administration.5 The 1997 guidelines on asthma management from the expert panel of the National Education and Prevention Program of the National Heart, Lung, and Blood Institute6 state emphatically that inhaled corticosteroids provide the most effective long-term control of asthma. The most cost-effective way of reducing acute asthma symptoms with antiinflammatory therapy is to follow a short course (1 week or less) of a systemic cor-

568

ticosteroid with chronic use of an inhaled corticosteroid.6

The Costs of Asthma The total cost of managing a chronic condition such as asthma can be broken down into direct, indirect, and intangible costs. The direct costs of asthma include charges for the use of health services, such as medical consultations, laboratory tests, spirometry, radiography, hospital inpatient stays, and emergency room visits, and the cost of the purchase of asthma drugs and aerosol delivery systems. Non-healthrelated direct costs include patients’ time spent traveling to and waiting for appointments. The indirect costs, such as lost productivity, and intangible ones, such as reduced quality of life, social costs, and psychological problems, are difficult to quantify, particularly when using secondary dam7 The total estimated cost of asthma in the United States was nearly $4.5 billion in 1985, with direct medical expenditures approaching $2.4 billion, or 53% of the total. The largest category of direct medical expenditures was inpatient hospitalizations ($1 billion). Indirect costs were estimated to exceed $2 billion. When the 1985 estimates were projected to 1990 dollars, the total cost of asthma in 1990 was estimated at $6.2 billion.7 Medicaid is a program of national health assistance for individuals and families with low incomes who are elderly, blind, or otherwise disabled and for members of families with dependent children. It is funded by the federal government and the states. Drug reimbursement is made on a retrospective, fee-for-service basis, with payments limited to the lower end of the usual and customary charge of

R. BALKRISHNAN ET AL.

the pharmacy that fills the prescription or to the pre-established Medicaid rate. The state of North Carolina pays for the comprehensive care of eligible patients, including their hospitalizations, emergency room visits, and physician visits. The state spends approximately $2800 per recipient annually for its estimated 1 million Medicaid recipients. Drug payments are approximately $220 per patient annually, less than the costs incurred for any other health-care service.8 Per-patient Medicaid costs are increasing nationwide. During the past decade, the proportion of hospitalizations for asthma increased among children who are covered by Medicaid or are uninsured,7 and Medicaid has recently been identified as the third-party payer with the highest rate of hospitalization for asthma.9,10 Improper drug therapy and improper use of drug therapy can cost Medicaid thousands of dollars per patient for hospitalizations, emergency room visits, and physician visits. Thus although the addition of inhaled corticosteroids to existing asthma therapy would increase Medicaid’s per-patient drug cost, the additional therapy could reduce total health care costs to Medicaid while improving patient outcomes. If this could be demonstrated through a costbenefit analysis, Medicaid could encourage widespread use of inhaled corticosteroids to improve outcomes and decrease per-patient costs of asthma treatment.

Pharmacoeconomic

Studies to Date

Few pharmacoeconomic studies have been conducted that assess the costs associated with the use of inhaled therapy for asthma, although there have been exhaustive studies assessing the overall costs of this disease. Studies that have addressed

this topic include a cost comparison of bronchodilator use, l l an economic evaluation of formoterol versus salmeterol inhaler use,12 and a cost assessment of albuterol inhaler use in certain Medicaid populations.13 Three studies of the costs of inhaled corticosteroid therapy were found in the literature, the most significant of which was conducted by Rutten-Van Molken et all4 to determine the costs and effects of combined bronchodilator and antiinflammatory therapy for asthma. Overall, the investigators found that the addition of inhaled corticosteroid therapy led to a small net increase in health care costs of $201 per patient per year. To reach net social savings, the increased productivity resulting from inhaled corticosteroid use would be expected to be greater than $42 per day. The authors concluded that the addition of a corticosteroid to be%-agonist therapy led to significant improvements in respiratory function and restricted-activity days that justified the relatively low increase in health care costs. An earlier prospective cohort study by the same research group15 compared the cost-effectiveness of inhaled beta,agonist therapy plus an inhaled corticosteroid with that of an inhaled beta,agonist plus placebo in 116 Dutch asthmatic children aged 7 to 16 years. The researchers calculated that there had been a 70% reduction in direct health care costs, a 55% reduction in indirect health care costs, and a 68% reduction in overall health care costs as a result of the introduction of corticosteroid therapy. Pererat6 conducted a cost-effectiveness study of inhaled corticosteroid use in 86 Sri Lankan children with asthma. A significant proportion of the patients showed improvement in clinical variables with inhaled corticosteroid therapy, confirming

CLINICAL THERAPEUTICS”

the efficacy of this therapy. The mean monthly cost per patient was reduced by 80% after the initiation of inhaled corticosteroid therapy. None of the aforementioned studies investigated the regularity of prescription refills, which could be an important factor in both patient outcomes and total health care costs.

Study Objectives and Perspective The objectives of the present study were: (1) to evaluate the impact of inhaled corticosteroids, in combination with existing therapy, on the use of health care resources, including hospitalizations, physician visits, and outpatient visits, by patients with moderate-to-severe asthma or asthma-related conditions in the North Carolina Medicaid population; and (2) to evaluate the impact of inhaled corticosteroids on total health care costs (ie, cost of drug therapy, physician and outpatient facility visits, and hospitalizations) per patient per year in that population. Cost-benefit analysis is used to compare two or more interventions and measure their costs and outcomes in monetary units. Because all outcomes are measured in a single unit (dollars), cost-benefit analysis can be used to compare various courses of action having different outcomes. It assumes that there are limited or finite resources to be expended and is typically used to choose the option with the greatest return. However, cost-benefit analysis cannot be used to measure nonmonetary outcomes such as quality of life. The outcome measure used in the present study was net benefit, which was calculated by subtracting total costs from total benefits.17 We took the perspective of a third-party payer (Medicaid).

570

MATERIALS

AND METHODS

Study Population The target population consisted of all North Carolina Medicaid recipients. Data were extracted from two large files in the North Carolina Medicaid database, one containing paid claims data on medical services provided during the period March 1993 to March 1996 for conditions with ICD-9 (International Classification of Diseases, 9th Revision’*) codes for asthma or an asthma-related condition and the second, the history file, containing paid claims records (including drug claims recorded with National Drug Committee [NDC]19 codes) for the same population. The two files were merged and two separate subset files created, one containing all paid claims (including drug claims) for the case group and the other containing all paid claims for the control group (groups defined below). Because names were blanked to protect patients’ identity, the unique patient identification number and service dates were used to link the files. The case population was defined as persons who were eligible to receive Medicaid for a continuous 3-year period from March 1993 to March 1996 and who: (1) had a paid claim for asthma (ICD-9 493.00) or an asthma-related condition (ICD-9 493 followed by two digits; for example, acute asthmatic bronchitis 493.90) or asthmatic complication classified under chronic obstructive pulmonary disease (COPD)* (ICD-9 491 followed by two digits; for example, acute and chronic

*These were patients whose claims records showed diagnoses of both asthma and asthma complications classified as COPD.

R. BALKRISHNAN ET AL.

asthmatic bronchitis COPD 491.21) before March 1995; (2) had started inhaled corticosteroid therapy between March 1994 and March 1995, as evidenced by drug claims; and (3) had continued inhaled corticosteroid therapy for at least 1 year from the start of therapy. Patients could have been taking other drugs for asthma during this period. The control population consisted of persons who were eligible to receive Medicaid for a continuous 3-year period from March 1993 to March 1996, who had been diagnosed with conditions having the same ICD-9 codes as the case patients, and who were receiving any asthma therapy except inhaled or oral corticosteroids for a continuous 2-year period. Study Design This retrospective cohort study covered the period March 1993 to March 1996. Its time frame extended from 1 year before to 1 year after the start of inhaled corticosteroid therapy for the case group and consisted of 2 continuous years without inhaled corticosteroid therapy for the control group. The case group included patients who began inhaled corticosteroid therapy between March 1994 and March 1995. Case patients who had not been using inhaled corticosteroid therapy for at least 1 year and those who were using inhaled corticosteroid therapy before March 1994 were excluded from the study. The control group consisted of patients who were using any treatment for asthma other than inhaled or oral corticosteroids for a continuous 2-year period between March 1993 and March 1996. Patients who had incomplete records (because of ineligibility for Medicaid for certain time periods) and patients taking corticosteroids for any

indication other than asthma were also excluded from the study. Resource utilization and economic variables for the relevant time period were retrieved from the database for use in two comparisons: (1) a comparison of costs and health care utilization before and after the start of inhaled corticosteroid therapy in the case group; and (2) a comparison of the costs and health care utilization of the case group with those of the control group. The resource-utilization variables included clinical procedures, outpatient facility visits (including emergency room visits), physician visits, hospitalizations, and drugs used. The cost variables included Medicaid reimbursement for clinical procedures, drug therapy, outpatient facility visits, physician visits, and hospitalizations. Operational Inclusion Criterion A problem associated with the data set used in this study was that claims information was missing for some recipients during certain periods. Since it was not possible to determine clearly whether a patient had ceased to be eligible or had not used health care services during these periods, an operational criterion for follow-up was used for each patient in the case and control groups. For inclusion, it was necessary that the patient have claims records for any health care service utilization for at least 6 months of each of the 2 years included in the study. The claims could be for prescriptions, non-drug-related health care services, or both and were not necessarily limited to asthma-related expenses. Additionally, it was necessary that each case patient have claims records for at least 2 months of each 6-month period in the year immedi-

571

CLINICAL THERAPEUTICS’

ately preceding or following the initiation of inhaled corticosteroid therapy (ie, at least two separate claims in 2 months within each 6-month period for four 6month periods). For the control group, there had to be at least 2 months of claims data for each 6-month period. For example, using this operational criterion, a case patient with exactly 6 months of claims data immediately preceding or following the start of inhaled corticosteroid therapy or a case patient with only one claim in each 6-month period would not be included in the study. This criterion was used because many patients require asthma medications or make use of health care services for their asthma only seasonally. Thus their use of health care services and asthma drugs is likely to be greater in one 6-month period than another. Use of the operational criterion is helpful in dealing with the problem of incomplete data but has the potential to create such problems as smaller sample sizes. This problem, which is discussed in a later section, was encountered in the present study.

Assumptions Certain assumptions were adopted because of the limitations of using a reimbursement database and a retrospective cohort study design. It was difficult to assess whether a patient had used inhaled corticosteroids or any other asthma drug appropriately. We could only check for regular drug-refill patterns and assume patients understood the use of inhalers. The severity of asthma had been recorded in the database by ICD-9 code; we assumed that the coding had been done correctly. It was also assumed that the outcomes studied were the result of therapy

572

and not of any other factor. Finally, we assumed there was no link between the claims data for medical services and the data on drug reimbursement except patient identification number and service date.

Demographic and Utilization Variables Variables used for the data analysis were categorized as demographic, utilization, and cost variables. Age and gender were the two demographic variables available in the North Carolina Medicaid database. For the purposes of the study, patient health care resource utilization variables from the database were grouped into disease-related and treatment-related variables. The presence of the disease was confirmed by checking whether the patient had received a diagnosis with an ICD-9 code for asthma before March 1, 1993. The presence of other diseases (comorbidities) was confirmed by examining patients’ first five secondary diagnosis fields, which were also recorded using ICD-9 codes. Clinical procedures performed and drugs dispensed had been recorded in the database using Current Procedural Terminology20 and NDC codes, respectively. The reimbursement data were used to establish the number of hospitalizations, clinic visits, and outpatient facility visits. Inhaler Refill Regularity Coeflcient A method devised by Suissa et a12i was used to measure the regularity of inhaler refills in the case group. The number of canisters of inhaled corticosteroids dispensed each month was obtained from each patient’s records. A profile score was calculated based on inhaled corticosteroid use in the 12 months prior to the index event date (the last date of available data

R. BALKRISHNAN ET AL

for the patient). The profile scores varied from 0 to 11; in general, a score below 5.5 indicated underuse, a score of 5.5 indicated constant regular use, and a score above 5.5 indicated overuse. The regularity coefficient was calculated using the formula (5.5 - deviation of score from 5.5)/5.5. Thus patients who had a regular pattern of inhaler refill received a score of 1, and all others received scores below 1.

Diagnosis CoefJicient A diagnosis coefficient was calculated to assess comparability of asthmatic conditions between case and control patients as follows. If a patient had reimbursements for only the ICD-9 codes 493.00 (asthma), 493.1 (allergic bronchial asthma), 493.2 (chronic obstructive asthma), 493.9 (acute asthmatic bronchitis), or 493.21 (obstructive asthma with status asthmaticus), we assumed that the severity of disease was moderate, and a coefficient of 1 was assigned. If the patient had a diagnosis of 491.20 (chronic asthmatic bronchitis classified under COPD) or 49 1.2 1 (acute and chronic asthmatic bronchitis classified under COPD), we assumed that the disease was moderate to severe, possibly complicated by COPD, and a coefficient of 2 was assigned. A single diagnosis was enough to determine the coefficient for the year; for example, one diagnosis of 491.21 gave the patient a diagnosis coefficient of 2 for the year.

utilization data for both the case and control groups. Since the follow-up period for each patient spanned more than 1 year, costs for the second year were deflated using the medical care component of the Consumer Price Index (CPI) as follows. The medical care component of the CPI changed by 5.470, 4.990, 3.990, and 3.0% in the years 1993, 1994, 1995, and 1996, respectively. 22 Since the study encompassed the period March 1993 to March 1996, the deflation factor was calculated as (0.75 * 5.4 + 1 * 4.9 + 1 . 3.9 + 0.25 * 3)/3 = 4.53%. Thus the costs in the second year of the study were deflated by a factor of 1.0453.

RESULTS

Demographic Differences Using the operational inclusion criterion, 180 case patients and 233 control patients were identified. Of these, only 38 patients (15 in the case group and 23 in the control group) had a diagnosis coefficient score of 2. Variations in baseline characteristics between the case and control groups are illustrated in the figure. There were significant differences between case and control patients in the patterns of age, sex, and number of drugs used at the end of the second year (P < 0.05). Most striking was the fact that the two groups had almost identical diagnoses and presence of comorbidities.

Cost Variables

Differences in Utilization

The cost variables used included reimbursements for drug therapy, hospitalizations, physician visits, and outpatient facility visits. These were tallied separately for the first and second years, as were the

Paired t tests were used to assess whether there were any significant differences in the number of hospitalizations, physician visits, or outpatient visits between the case and control groups at base-

573

CLINICAL THERAPEUTICS”

100 rn

50

B

40

E

m

Case

0

Control

n" 5 30

1 Bg 20 f 0 10 $ I no

l-15

il-im.a 1 31-45 46-60

>60

Diagnosis Codes*

Age 701

Male

l.k 4

Female

Sex

2

60

g

70

.c 5 25 a 5 20

.; 60 5 a

50

ti 40 $

30

5 20 P Q) IO a 0

30

8 S 3

10

15

& a

5 0

Comorbidities Present

%I

No. of Drugs Used in Year 1

0

12

No. of Drugs Used in Year 2

Figure. Distribution of baseline characteristics for case (n = 180) and control (n = 233) groups (P c 0.05 in all catagories, except Diagnosis Codes and Comorbidities Present). *Diagnosis code 1 represents asthma; diagnosis code 2 represents asthmatic complications classified under chronic obstructive pulmonary disease. line, since the data were often normally distributed. In the first year, the number of physician visits was significantly lower in the control group; however, in the second year, the number of physician visits

574

decreased in the case group and increased in the control group, so the difference was no longer significant. The number of outpatient visits was significantly lower in the control group than in the

R. BALKRISHNAN

ET AL

case group during the first year, but the opposite held true in the second year. Outpatient visits were significantly higher in the control group than in the case group. No significant betweengroup differences in hospitalizations were noted in the first year; in the second year, hospitalization rates were significantly higher in the control group compared with the case group (P c 0.05). Paired t tests were also used to compare differences in health care service utilization in the 2 years. The case group had decreases of 50%, 26%, and 15% in hospital visits, outpatient visits, and physician visits, respectively (Table I), at the end of the second year. Significant reductions in health care utilization from the first to the second year after the introduction of inhaled corticosteroid therapy were

evidenced by decreases in hospital visits (P < O.OOOl), outpatient visits (P < 0.005), and physician visits (P < 0.005). Across the 2 years, the control group had increases of 23%, 36%, and 10% in hospital visits (P c O.Ol), outpatient visits (P < OBOS), and physician visits (not significant), respectively (Table II). Differences in Health Care Costs The cost benefit of inhaled corticosteroid therapy was estimated using loglinear multiple regression analysis (Table III). Regression analysis was used to adjust for the effects of possible confounders such as age, sex, and the presence of comorbidity. Since the distribution of total costs per asthmatic patient per month for the year was found to be skewed, the nat-

Table I. Difference in the use of health care services per patient per month in the case group from year 1 to year 2 (n = 180). Year 1 (Mean f SD)

Year 2 (Mean f SD)

Difference

Type of Service Hospital visits Outpatient visits Physician visits

0.34 f 0.50 0.38 f 0.48 1.05 f 0.81

0.17 f 0.28 0.28 f 0.37 0.89 zt 0.77

-0.17 -0.10 -0.16

(Mean)

P
Table II. Difference in the use of health care services per patient per month in the control group from year 1 to year 2 (n = 233). Year 1 (Mean f STD)

Year 2 (Mean f STD)

Difference

Type of Service

(Mean)

P

Hospital visits Outpatient visits Physician visits

0.26 f 0.47 0.25 -c 0.38 0.88 f 0.62

0.32 f 0.50 0.34 f 0.49 0.97 f 0.70

0.06 0.09 0.09


NS = not significant,

575

CLINICAL THERAPEUTICS”

Table III. Results of multiple regression analysis (N = 826). The dependent variable was LTPAID (natural logarithm of total costs per asthmatic patient per month for Medicaid). F ratio: 83.087 (P < 0.0001); adjusted R2:0.4432.

Variable Constant Age Age squared Sex Diagnosis coefficient Comorbidity Year Group Year-Group

Parameter Estimate

SE

T for H,, (Parameter = 0)

3.877 0.002 0.000 -0.049 2.09 0.242 0.094 0.097 -0.270

0.077 0.004 0.000 0.048 0.104 0.057 0.063 0.067 0.095

50.417 0.573 2.353 -1.021 20.061 4.227 1.516 1.439 -2.852

Dummy Variable Assignment Year [l(Year2), O(Yearl)] Group [ l(Case), O(Control)] Diagnosis coeffkient [ l(Coeffkient 2), O(Coeffkient Sex [l(Male), O(Female)] Comorbidity [l(F’resent), O(Absent)] NS = not significant.

ural logarithm of this variable was used as the dependent variable. Approximately 45% of the regression could be explained by the variables included. The variable for “Age squared’ was found to be significantly correlated with total health care costs per patient per month, but the magnitude of the effect was small. A change from diagnosis coefficient 1 to diagnosis coefficient 2 produced a significant, nearly sevenfold increase in total health care costs per patient per month (P < 0.0001). This probably occurred because the patient group with a diagnosis coefficient of 2 included some who were frequently hospitalized with complications of COPD. Comorbidity was significantly positively correlated with total health care

576

P>l 0.0001 NS

0.0189 NS 0.0001 0.0001 NS NS 0.0045

l)]

costs per asthmatic patient per month (P < 0.0001). The presence of comorbidity was associated with an average increase in health care costs of nearly 28% per patient per month. The parameter estimate of “Group” was not significantly correlated with total health care costs per asthmatic patient per month. Thus the case and control groups were comparable in the first year. The parameter estimate of the “YearGroup” interaction was found to be negatively correlated, which is to be expected when the therapy is cost beneficial, and was significantly correlated with total health care costs per asthmatic patient per month. The regression was checked for multicollinearity, which was not found to be a problem.

R. BALKRISHNAN ET AL.

Cost Benefit of the Introduction of Inhaled Corticosteroid Therapy The cost benefit of inhaled corticosteroid therapy for the case group was calculated from the regression analysis, which controlled for such factors as age, sex, comorbidity, and diagnoses. The parameter estimate of the “YearGroup” interaction is the net benefit of the therapy after adjusting for other influences. The parameter estimate was found to be -0.267 (P < 0.005). Therefore, the net benefit of the therapy was antilogarithm (-0.267) - 1, which was found to be -0.2366 (23.66%). Therefore, we concluded that there was a decrease of 23.66% in total health care costs per patient per month after the introduction of inhaled corticosteroid therapy. Difference in Costs with Regularity of Inhaler Refills Another regression analysis was performed on the case group to measure the difference regularity of inhaler refills made on the natural logarithm of total health care costs per asthmatic patient per

month (Table IV). The parameter estimate of inhaler refill regularity was not significantly correlated with the total cost per asthmatic patient per month. Therefore, we concluded that regularity of steroid inhaler refills was not associated with a reduction in health care costs per asthmatic patient in this population.

DISCUSSION This study is one of the first retrospective cost analyses of inhaled corticosteroid use to employ Medicaid claims data. The three studies mentioned previously’4-16 were cost-effectiveness analyses in which effectiveness was measured in “quality” units such as symptom-free days gained. Unlike these studies, we measured both costs and effectiveness in the same monetary unit. Results of this and similar studies could help Medicaid make clearer decisions about the allocation of financial resources, although it would not be possible to come to definitive conclusions about improvements in clinical end points such as symptom-free days gained. The advantage of the study design used here is that unlike the clinically controlled

Table IV. Regression analysis for inhaler refill regularity (P < 0.0001); adjusted R2: 0.1536.

Variable Constant Age Age squared Sex Comorbidity Regularity

(N = 180). F ratio:

Parameter Estimate

SE

T for H, (Parameter = 0)

3.805 0.018 -0.000 -0.045 0.233 -0.256

0.187 0.010 0.000 0.106 0.129 0.255

20.339 1.867 -0.530 -0.427 1.815 -1.003

6.314

P>t

0.0001 0.063 NS NS 0.07 1 NS

NS = not significant.

577

CLINICAL THERAPEUTICS”

environment of a randomized controlled trial, it reflects real-world drug usage. Disadvantages of our study design include selection bias because of the inability to select patients randomly and problems of missing data because of the reliance on recorded claims data.

Study Limitations The limitations of this study stem primarily from the data set used. First, the final sample size (180 cases, 233 controls) was modest. Even though the initial data set contained a large number of patients, the sample size was reduced because many patients had missing claims records for long periods and therefore did not meet the operational inclusion criterion. The effects of the small sample size were evident in the parameter estimates of the multiple regression analysis for inhaler refill regularity. At baseline, many characteristics of the case and control groups were found to be comparable, but not age and sex. The control group was older than the case group and contained a larger number of females. Also, the control group contained 18 patients whose health care costs averaged $1000 per month or more, compared with only 5 such patients in the case group. Thus the distribution of total health care costs was skewed. The effect of this was that health care expenses in the first year were significantly higher in the control group than in the case group. In part, these higher costs may also have been the result of the older control patients having more comorbidity and the fact that females in general have greater health care expenses than males. The information available in the data set posed other problems. For example, it

578

was not possible to determine exactly why a patient was hospitalized or why a particular drug was prescribed, nor was it possible to quantify adverse drug effects and costs. The only measure of severity was the ICD-9 code assigned. The boundaries between ICD-9 codes for asthma are not very distinct, and although the diagnosis coefficients calculated using these codes may not have been accurate measures of the actual severity of a patient’s condition, they are a measure of how similar asthmatic conditions were across the two groups. Because health care utilization patterns were not very different between patients with these ICD-9 codes, it was assumed for the purposes of calculating the diagnosis coefficient that there were no distinct differences between ICD-9 codes 493.00, 493.21, and 493.90. It was also assumed that the ICD-9 coding was done accurately. Comorbidity was recorded as a dummy variable (present or absent). Another limitation of the study was the inability to distinguish between costs in patients with differing comorbidities. Finally, there may have been a selection bias because we were not able to assign patients randomly to the case and control groups.

CONCLUSIONS We found that the introduction of inhaled corticosteroid therapy has a cost benefit for Medicaid and can bring about significant reductions in the use of health care services and costs. Greater use of inhaled corticosteroid therapy should be advocated to improve medical outcomes in patients with moderate-to-severe asthma. This could reduce unnecessary hospitalizations, outpatient and clinic visits, and

R. BALKRISHNAN

ET AL.

costs associated with improperly treated asthma in these patients. Future studies should focus on larger sample sizes to make these results more generalizable. This would require the pooling of similar data from other states’ Medicaid databases. Studies are also needed to examine more precisely such factors as asthma severity, comorbidities, and the direct effects of inhaled corticosteroid therapy. Since our study was conducted, newer drug treatments for asthma, such as the leukotriene antagonists, and more potent inhaled corticosteroids have emerged. The cost benefits and outcomes associated with these newer therapies need to be studied in detail and compared with current therapies.

ACKNOWLEDGMENTS Financial support for this paper was provided by Glaxo Wellcome Inc., Research Triangle Park, North Carolina. The authors would also like to thank Dr. Teresa Kauf, Dr. Betsy Sleath, and Dr. Ya-Chen Tina Shih from the University of North Carolina for their input and reviews. Address correspondence to: Rajesh Balkrishnan, MS (Pharm), Division of Pharmaceutical Policy and Evaluative Sciences, School of Pharmacy, University of North Carolina at Chapel Hill, CB #7360, Beard Hall, Chapel Hill, NC 27599-7360.

REFERENCES 1. Pearce N, Beasley R, Burgess C, et al. Asthma Epidemiology, Principles and Methods. New York: Oxford University

Press; 1998:75-79.

2. Mellis CM, Peat JK, Woolcock AJ. The cost of asthma: Can it be reduced? PharmacoEconomics. 1993;3:205-219. 3. Bauman A, Mitchell CA, Henry RL, et al. Asthma morbidity in Australia: An epidemiological study. Med JAust. 1992; 156: 827-831. 4. Co1 N, Fanale J, Kronholm P. The role of medication non-compliance and adverse drug reactions in the hospitalizations of the elderly. Arch Intern Med. 1990;150: 841-845. 5. Fong PM, Sinclair DE. Inhalation devices for asthma. Can Fam Phys. 1993;39: 2377-2382. 6. Expert Panel Report II: Guidelines for Diagnosis and Management of Asthma. Bethesda, Md: National Heart, Lung, and Blood Institute, 1997. 7. Weiss KB, Gergen PJ, Hodgson TA. An economic evaluation of asthma in the United States. NEJM. 1992;326:862-866. 8. Pharmaceutical Benefits Under the State Medical Assistance Programs. Reston, Va: National Pharmaceutical Council, 1995: 79-226,42843 1. 9. Wissow LS, Gittelsohn AM, Szko M, et al. Poverty, race and hospitalization for childhood asthma. Am J Public Health. 1988:78:777-782. 10. Carr W, Zeitel L, Weiss KB. Variations in asthma hospitalizations and deaths in New York City. Am J Public Health. 1992;82: 59-65. 11. Nightingale CH. Cost comparison of beta 2-agonist bronchodilators used in the treatment of asthma. Pharmacotherapy. 1995; 15:677-681.

579

CLINICAL THERAPEUTICS”

12. Sculpher MJ, Buxton MJ. Episode free day as a composite measure of cost effectiveness. PharmacoEconomics. 1993;4: 345-352. 13. Tierce JC, Meller W, Berlow B, et al. Assessing the cost of albuterol inhalers in the Michigan and California Medicaid programs. Clin Ther 1989;11:53-61. 14. Rutten-Van Molken MPMH, Doorslaer EKAV, Jansen MCC, et al. Costs and effects of inhaled corticosteroids and bronchodilators in asthma and chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 1995;151:975-982. 15. Rutten-Van Molken MPMH, Doorslaer EKAV, Jansen MCC, et al. Cost effectiveness of inhaled corticosteroid plus bronchodilator versus bronchodilator monotherapy in children with asthma. PhannacoEconomics. 1993;4:257-270. 16. Perera BJC. Efficacy and costs of inhaled steroids in asthma in a developing country. Arch Dis Child. 1995;72:312-316.

580

17. Eisenberg JM. Clinical economics: A guide to the economic analysis of clinical practices. JAMA. 1989;262:2879-2885.

18. International Classification of Diseases, 9th Revision. Los Angeles, Calif: PMIC; 1995.

19. 1996 Drug Topics@ Red Book. Montvale, NJ: Medical Economics: 1996.

20. American Medical Association. CPT 1995: Physicians’ Current Procedural Terminology. Salt Lake City, Utah: Medicode Inc.; 1995.

21. Suissa S, Blais L, Ernst P Patterns of increasing beta-agonist use and the risk of fatal and near fatal asthma. Eur Respir J. 1994;7:1602-1609.

22. Bureau of Labor Statistics. Available at: Consumer Price Index summary. http://stats.bls.gov. Accessed July 15, 1997.