| |
Medicaid Utilization of IFN/RBV for HCV Therapy
|
| |
| |
"Factors associated with the initiation of alpha-interferon treatment in Medicaid patients diagnosed with hepatitis C"
Journal of Viral Hepatitis- March 2005
J. S. Markowitz 1 , E. M. Gutterman 1 , D. Hodes 2 and W. Klaskala 2
1Health Data Analytics, Princeton, NJ, USA; and 2Roche Laboratories Inc., Nutley, NJ, USA
AUTHOR CONCLUSION
Low rates of HCV treatment in this Medicaid-insured population may reflect systemic obstacles or physician's disinclination to treat, as well as treatment appropriateness associated with patient selection. This study found no evidence that pre-existing comorbid conditions, included in FDA cautionary statements, such as depression, were acting as an obstacle to treatment. Nevertheless, substantial rates of comorbid disease, ranging from psychiatric illness to diabetes to pulmonary conditions and high use of concomitant drug therapy, highlight the challenges of integrating HCV treatment into comprehensive medical care in similar patient populations.
Summary-
We aimed to determine rates of treatment with alpha-interferon medication in patients diagnosed with hepatitis C virus (HCV), to ascertain the prevalence of selected conditions that could influence initiation of interferon treatment, and to examine the association between the presence of these conditions and interferon treatment. A nested case-control design was used in California Medicaid (Medi-Cal) claims data covering the period from 1 January 1996 to 30 June 2002. Interferon-treated cases and non-treated controls were selected in a 1 : 2 ratio that matched the length of the observation period and year of index HCV diagnosis. Predictor variables examined in bivariate and multivariate analyses included demographics, substance abuse and dependence, psychotropic drug use, selected chronic conditions and medical utilization. The proportion of eligible subjects diagnosed with HCV and treated with interferon ranged from 10.7 to 13.9%. There were 529 treated cases that met the eligibility criteria and 1058 non-treated HCV patients selected as controls. Multivariate factors that increased the likelihood of treatment were a liver biopsy, a diagnosis of mild liver disease, a diagnosis of psoriasis, antidepressant use and classification of race/ethnicity as 'other'. A decreased likelihood of treatment was linked to age >=65 years, a diagnosis of kidney disease, one to four emergency visits and five or more emergency visits. The proportion of patients receiving interferon treatment in the Medi-Cal-insured population was low compared with published rates in HCV patients in other general medical settings. The diverse factors linked to initiation of HCV therapy raise compelling questions for further research.
AUTHOR DISCUSSION
HCV treatment in the Medi-Cal patient population
Among the HCV patients who met study criteria (at least two HCV claims and a pre-diagnosis period of 365 days), the proportion of patients who received HCV treatment was 10.7%. This proportion was raised to 13.9% when patients who received HCV therapy in the 365 days prior to the first HCV claim were included in the analysis.
Compared with published rates, the proportion of patients treated for HCV was low. Age and sex differences in the samples are unlikely to explain the disparity in rates. One could argue that the period studied was characterized by marked changes in the treatment of HCV prompting physicians to hold off on treatment until a better therapy was approved. However, published studies covered overlapping time frames. One study of a large New England population insured through indemnity plans or managed care in 1995-97 identified a rate of IFN- alpha treatment of 51.3% (98/191) [17]. This claim-based study also required patients to have at least two claims for HCV as evidence of an HCV diagnosis and did not consider liver enzyme levels and histology reports that routinely influence treatment decisions but are unavailable in claims data [17]. Another study, based on chart reviews in primary care settings for adults, found an IFN treatment rate of 29% [16]. Within the Veteran's Administration (VA), there was an IFN- alpha treatment rate of 71% (145 of 206 consecutive patients) in Minneapolis [15] and a 32% rate of eligibility for treatment in Durham, North Carolina [14]. In the latter study, lack of eligibility was based on severe mental illness, hazardous alcohol consumption, current drug abuse, decompensated cirrhosis, dementia, terminal illness, diabetic ketoacidosis, severe cardiac or pulmonary disease or homelessness.
Specifically, the rate of IFN- alpha treatment in this Medi-Cal sample was less than half the lowest rate (29%) identified in published studies [16]. Although we cannot rule out the possibility that there were larger numbers of persons who did not meet recommended criteria for treatment (e.g. individuals with normal liver enzyme levels that did not require treatment or individuals with psychiatric, physical, or social situations that made them too unstable to receive treatment [2]), the variability in prescribing rates of IFN- alpha to HCV patients merits further investigation. Additional research could address the role of type of insurance, level of prescription coverage and physician variables (e.g. specialty and access to diagnostic procedures).
What factors increase the likelihood of treatment?
The factors that increased the likelihood of treatment in the final multivariate model were: the occurrence of liver biopsy, a diagnosis of mild liver disease, a diagnosis of psoriasis, antidepressant use and classification of race/ethnicity as 'other'. Factors that decreased the likelihood of treatment were: being of age >=65, a diagnosis of kidney disease and multiple emergency visits.
These results offer insights about the role of liver biopsy, antidepressant use and age on treatment status, which are addressed in subsequent sections. In general, physicians appear to prescribe IFN to patients despite the existence of conditions identified in FDA warnings. Such treatment may be consistent with good medical practice provided that these patients are followed more closely. In this study, a wide range of conditions that could be included in the disease categories described in the warnings ('neuropsychiatric, autoimmune, ischemic, and infectious disorders') were not significantly associated with the decision to treat in adjusted analyses. These conditions included AIDS/HIV, depression, diabetes, ischaemic heart disease, lupus and rheumatoid disorder. Nevertheless, a diagnosis of kidney disease was linked to a lower likelihood of treatment, perhaps because kidney dysfunction interferes with the metabolism of drug treatment. Furthermore, despite indications that IFN treatment may exacerbate psoriasis, patients with psoriasis had a significantly higher likelihood of treatment [8-10]. This finding, along with higher proportions of rheumatoid disease, peptic ulcer disorder and ischaemic heart disease in cases compared with controls (albeit at non-significant levels), may indicate that patients under the close care of physicians for certain conditions are more likely to receive IFN treatment for HCV.
The impact of occurrence of a liver biopsy on HCV treatment
The results suggest that patients likely to receive a liver biopsy are also likely to receive IFN- alpha treatment for HCV. This result is not surprising given consensus guidelines for the management of HCV, which describe liver biopsy as 'a unique source of information on fibrosis and assessment of histology, that may allow patients to make more informed decisions in initiating or postponing antiviral treatment' [30]. Nevertheless, two scenarios are possible: the results of the liver biopsy may drive the decision to treat with IFN or the propensity to treat with IFN may drive the decision to perform a liver biopsy. Although most physicians in one integrated health delivery system indicated that they recommended liver biopsies for all their HCV patients who have elevated alanine aminotransferase (ALT) levels, the proportion of documented liver biopsies in the charts of such patients was only 36% [16]. Consequently, further studies are needed to examine physician factors that predict rates of liver biopsy in HCV patients. Moreover, additional research could address the role of liver biopsy in patients' ability to make informed choices about HCV treatment and their readiness to undergo IFN- alpha therapy.
IFN-treated patients were more likely to use antidepressants relative to controls
The 42% greater likelihood of antidepressant use in IFN-treated patients relative to controls may suggest one of the following explanations. First, the rates of depression and/or other conditions treated with antidepressants were higher in IFN-treated patients. Based on claims with a diagnosis of depression, rates of depression were elevated in cases although not sufficiently higher to account for disparities in levels of antidepressant prescribing. Secondly, higher rates of antidepressant prescribing among cases could indicate that physicians caring for these patients were more likely to provide therapy for both depression and HCV. However, in this sample, there was no evidence that physicians were more likely to prescribe antianxiety drugs or antipsychotics. Thirdly, physicians planning to treat with IFN could have initiated use of antidepressants as a prophylactic strategy [31,32] to reduce the risk of depression associated with IFN [33,34]. Although there is some evidence that pretreatment with antidepressants may mitigate the occurrence of IFN-induced depression [32], the extent of prescribing antidepressants for this purpose has not been studied. Future research could examine the timing of the start of antidepressant treatment as a way to determine the type of connection between antidepressant treatment, HCV infection and IFN- alpha treatment.
The lower likelihood of IFN treatment in older individuals
The risks of IFN- alpha treatment in the elderly, in terms of possible adverse events or drug-drug interactions, combined with low perceived benefit with respect to increased economic productivity, may help explain the lower rates of prescribing in the >=65-year-old group. Nevertheless, increased rates of sustained viral response with new forms of IFN- alpha therapy [4], as well as limited research suggesting a more rapid progression of liver disease [35,36] and higher rates of hepatocellular carcinoma [37] in older patients who contract HCV may suggest the need for more attention to the benefits of IFN- alpha therapy in this older group of patients.
Study limitations
The rates of IFN- alpha therapy in this HCV sample need to be interpreted in the context of the requirement to have at least two claims indicating a diagnosis of HCV. As a result of uncertainty regarding a diagnosis in the excluded patients, calculation of their rates of IFN- alpha treatment would be difficult to evaluate. The study sample size offered adequate power to detect group differences between cases and controls that were in the 10% range in categorical independent variables given an alpha level of 0.05 and a two-tailed test of significance [38]. Consequently, larger differences were likely to be detected while smaller ones could be missed.
Inclusion criteria for the HCV diagnostic claims encompassed both acute and chronic HCV. However, acute HCV infection is rarely diagnosed and, in addition, acute infection becomes chronic in most individuals [39]. Current consensus guidelines only address use of IFN therapy in chronic HCV, although there is growing evidence that treatment of acute HCV with IFN may be beneficial [40]. As a result of the lack of laboratory findings in claims data, there was no information on the presence of elevated liver enzymes. Nevertheless, guidelines for the management of HCV consider histological evidence of disease progression as well as serum aminotransferase levels in recommending which patients should receive IFN- alpha treatment [30].
Claims data also reflect prescriptions filled and there is no information on use or compliance. In addition, diagnostic information on claims data may underreport patients with chronic conditions, particularly in patients with multiple conditions requiring more intensive follow-up [41,42]. Finally, Medicaid data reflect the health utilization patterns of an indigent population and may not be generalizable to other population sectors. However, the nature of the data will also increase access to HCV cases found in economically deprived groups including intravenous drug users and individuals with chronic illnesses.
RESULTS
Determination of the sample
There were 10 269 patients with one or more HCV claims in the selected Medi-Cal sample. After excluding patients who lacked a pre-diagnosis period of 365 days and at least two HCV claims, this number was reduced to 5135 eligible HCV patients. The assignment of treatment status was the next step. Although there were 715 HCV patients with IFN- alpha claims, 529 patients remained after exclusion of patients who received IFN in the 365 days prior to the first HCV claim. These 529 patients were designated as cases. With respect to identification of controls, there were 4420 patients who had no claims for IFN- alpha in the 5135 eligible HCV patients. Following the study protocol, non-treated HCV patients were matched to the 529 cases, with a 2 : 1 ratio, resulting in the selection of 1058 control patients.
Rates of IFN treatment
Among HCV patients who fulfilled the study criteria (at least two HCV claims and a pre-diagnosis period of 365 days), the proportion of patients treated with IFN- alpha was 13.9% (715/5135). After the exclusion of patients who received IFN- alpha in the 365 days prior to the first HCV claim, the proportion was further reduced to 10.7% (529/4949).
Characteristics of IFN-treated and non-treated patients
Following HCV diagnosis, the 529 cases and 1058 controls were followed up for an average of 263.6 days (SD 318.6). Because all cases and controls had 1 year (365 days) of data prior to diagnosis, total follow-up during the observation period was a mean of 628.6 days.
Bivariate correlates of IFN treatment status
Demographics
In the distribution by age, 95.4% of cases were between ages 19 and 64 years, in contrast to 88.3% of controls (see Table 1). Consequently, compared with controls, cases had significantly lower proportions of patients in the youngest and oldest age groups as reflected by OR of 0.30 (95% CI 0.10-0.87) and 0.40 (95% CI 0.24-0.67), respectively. Nevertheless, the mean ages for cases (45.7 years) and controls (46.5 years) were not significantly different (P = 0.23). With respect to sex, the distribution of cases and controls was similar with males comprising 52.1% of cases and 51.5% of controls. In addition, Caucasians comprised similar proportions of each group with 55.3% in cases and 54.6% in controls. However, when compared with Caucasians, patients classified as 'other' had a higher likelihood of IFN- alpha treatment (OR 1.37, 95% CI 1.03-1.83), while patients whose race/ethnicity was 'missing' had a lower likelihood of treatment (OR 0.66, 95% CI 0.47-0.93). In contrast, there were no significant differences in the likelihood of treatment in patients who were Black (OR 0.83, 95% CI 0.58-1.19) or Hispanic (OR 1.06, 95% CI 0.66-1.72) when these patient subgroups were compared with Caucasians.
Substance abuse and dependence
The ORs for abuse or dependence of alcohol, opioids, or other drugs were 0.72 (95% CI 0.49-1.07), 0.83 (95% CI 0.55-1.24) and 1.00 (95% 0.78-1.29), respectively, reflecting no significant differences between cases and controls. In cases and controls, the respective rates of abuse or dependence were 7.0%vs 9.5% for alcohol and 6.8%vs 8.1% for opioids, while the rate of abuse or dependence of other drugs was the same (21.3%).
Use of psychotropic drugs
The proportions using antidepressant drugs were 57.2% in cases compared with 47.1% in controls with cases significantly more likely to receive treatment with antidepressant medications during the observation period (OR 1.50, 95% CI 1.22-1.85). In contrast, the two groups were similar in their use of antipsychotics (23.4% in cases vs 24.1% in controls) and antianxiety drugs (31.3% in cases vs 28.8% in controls).
Selected chronic conditions
When cases and controls were combined, the most prevalent chronic conditions identified in this sample were: chronic pulmonary disease (32.0%), mild liver disease (27.4%), depression (22.2%), diabetes (19.1%), ischaemic heart disease (11.9%) and peptic ulcer disorder (10.3%). Other chronic conditions with a prevalence of at least 5% included: congestive heart failure (6.1%), cerebrovascular disease (6.0%), malignancy (6.0%), rheumatoid disease (5.9%) and human immunodeficiency virus/acquired immunodeficiency syndrome (HIV/AIDS) (5.6%).
When cases and controls were compared on the 17 conditions included in the Charlson comorbidity index, only three conditions emerged as significantly related to treatment status. Compared with controls, cases were less likely to have diabetes with chronic complications (OR 0.45, 95% CI 0.23-0.87) and renal disease (OR 0.38, 95% CI 0.19-0.76). However, cases were more likely than controls to have mild liver disease (OR 2.64, 95% CI 2.10-3.32).
Both groups were also compared on the prevalence of a number of chronic conditions that may be triggered or exacerbated by IFN treatment. Some of these conditions, chronic pulmonary disease, diabetes, myocardial infarction, renal disease and rheumatoid disorder, are part of the comorbidity index and thus already considered. Additional conditions in this category included depression, haemolytic anaemia, ischaemic heart disease, organ transplant, and automimmune diseases, such as Crohn's disease, lupus, multiple sclerosis and psoriasis. Among the additional conditions, depression (OR 1.30; 95% CI 1.02-1.66) and psoriasis (OR 2.70; 95% CI 1.13-6.46) were increased in cases. Other conditions were not significantly associated with the likelihood of treatment in comparisons between cases and controls. In cases and controls, the rates for depression were 25.3%vs 20.6%, while the rates for psoriasis were 2.3%vs 0.9%.
Medical utilization
Compared with controls, cases were more likely to have multiple physician office visits and an increased number of surgical procedures (excluding liver biopsy). For example, cases had an elevated risk of having thirteen or more physician office visits (OR 1.52, 95% CI 1.18-1.96), as well as a greater likelihood of one to five surgical procedures (OR 1.51, 95% CI 1.16-1.96) or six or more surgical procedures (OR 1.86, 95% CI 1.45-2.38). In addition, cases were far more likely to have had a liver biopsy during the observation period, as reflected by an odds ratio of 16.26 (95% CI 12.14-21.77). On the contrary, cases were less likely relative to controls to have a hospital claim (OR 0.36, 95% CI 0.17-0.78) or five or more emergency visits (OR 0.76, 95% CI 0.59-0.98).
Multivariate correlates of IFN treatment status
The first summary model included all dimensions with significant effects in domain-specific multivariate models that adjusted for demographic factors (domain-specific results not shown) (see Table 2). This model of selected variables, which also incorporated demographic factors, included alcohol abuse or dependence, antidepressant use, diabetes with complications, mild liver disease, moderate/severe liver disease, kidney disease, psoriasis, one to four emergency visits, five or more emergency visits and liver biopsy.
Three factors increased the likelihood of HCV treatment: the occurrence of a liver biopsy [adjusted odds ratio (AOR) 15.56, 95% CI 11.34-21.35], the presence of mild liver disease (AOR 1.66, 95% CI 1.23-2.24) and antidepressant use (AOR 1.42, 95% CI 1.09-1.85). In contrast, two factors decreased the likelihood of HCV treatment: a diagnosis of kidney disease (AOR 0.23, 95% CI 0.09-0.57) and the occurrence of five or more emergency visits (AOR 0.69, 95% CI 0.49-0.97). In addition, among demographic factors, an age of >=65 years decreased the likelihood of treatment (AOR 0.35, 95% CI 0.19-0.65) while a classification of race/ethnicity as 'other' increased the occurrence of treatment (AOR 1.97, 95% CI 1.40-2.79).
A second multivariate model used stepwise backwards elimination to identify the 'best' predictors of treatment status (see Table 2). Nine variables that were significant in the initial model were also significant predictors of HCV treatment in the second reduced summary model. In addition, two variables, a diagnosis of psoriasis and one to four emergency visits, that just missed significance in the initial models became significant in the final model.
INTRODUCTION
Consensus on treating chronic HCV infection [1-3] has been influenced by the increasing efficacy of therapies for HCV [4,5] coupled with the serious risk of chronic liver disease when no treatment is given. Nevertheless, HCV treatment also poses health risks. Food and Drug Administration (FDA) product information indicates that alpha-interferon (IFN- alpha ) treatment can lead to serious neuropsychiatric, autoimmune, ischaemic and infectious disorders [6]. Well-documented side-effects of treatment include depression and neutropenia [4,7], while rates of anaemia are elevated when IFN- alpha is combined with ribavirin [4]. Furthermore, case reports have associated IFN treatment with the induction or exacerbation of psoriasis [8-10], Sjogren's syndrome [11], rheumatoid arthritis [12] and hypothyroidism [13]. Consequently, individuals with these chronic conditions require additional clinical attention when treated with IFN- alpha .
To date, there has been little systematic study of the rates of treatment in populations with HCV and the prevalence of pre-existing conditions that may influence HCV treatment [14-17]. Because patients and physicians participating in clinical trials must meet specific critieria, treatment patterns in the general population cannot be inferred from such studies [18,19].
The study objectives were to: (i) determine the prevalence of HCV treatment (IFN- alpha based with or without ribavirin) in a Medicaid population; (ii) identify and characterize the conditions and factors (e.g. demographic and medical care utilization variables) that may influence the initiation of HCV treatment; and (iii) examine the association between the presence of the aforementioned conditions/factors and HCV treatment status. Claims data were used to identify 'newly' diagnosed HCV patients, who are defined as patients having no HCV-related claims for a period of 12 months.
This analysis will contribute to a better understanding of dimensions related to HCV treatment in patients diagnosed with hepatitis C in a large claims database spanning a 6-year period beginning 1 January 1996. Furthermore, based on its comprehensive focus on the chronic conditions that can be aggravated by IFN- alpha treatment, it is the first study to determine rates of these conditions in a Medicaid patient population and to evaluate systematically whether presence of these conditions influences initiation of HCV therapy following HCV diagnosis.
METHODS
Design
Using a nested case-control design, this retrospective analysis was conducted in a 20% random sample of approximately 2.08 million individuals covered by the California Medicaid (Medi-Cal) programme in the period between 1 January 1996 and 30 June 2002. The data consisted of pharmaceutical, medical and procedural claims linked at the individual patient level.
Identification of the sample
HCV diagnosis
All patients included in the study had a primary or secondary diagnosis of HCV infection based on codes identified in the International Classification of Diseases (9th Revision, Clinical Modification; CD-9-CM) [ICD-9: 070.51, 070.54] [20]. In addition, patients had to have at least two claims indicating a diagnosis of HCV; however, the first claim was used as the index HCV claim and served to demarcate the pre- and post-HCV diagnosis periods.
Pre-diagnosis period
Eligible subjects were required to have continuous enrollment in Medi-Cal for at least 12 months prior to the first HCV diagnosis. During this period, subjects could not have any medical claims with a primary or secondary diagnosis of HCV. Subjects were also excluded from the study if they had any prescription claims for IFN drugs during the pre-HCV diagnosis period.
Post-diagnosis period
For individuals prescribed any of the HCV target IFN- alpha -based drugs, such as Intron A (Schering-Plough, Kenilworth, NJ, USA), Roferon-A (Roche Pharmaceuticals, Nutley, NJ, USA), or Pegintron (Schering-Plough, Kenilworth, NJ, USA), the post-diagnosis period ended on the date of this drug claim. For individuals not prescribed any of these drugs, the length of the post-diagnosis period was equal to the post-diagnosis period of the matched case. Thus, data were truncated for controls whose follow-up exceeded matched cases.
Case-control classsification
All eligible patients who had at least one claim for IFN- alpha medication were designated as cases. Potential controls consisted of all eligible HCV patients who had no claims for IFN- alpha medication. Because potential controls greatly outnumbered cases, frequency matching [21] was used to reduce the effect of imbalances in the number of cases and controls. In addition, cases and controls were matched on the year of the index HCV claim and on the duration of follow-up [22]. In this manner, cases were compared with the subset of the cohort members that had a similar risk of exposure to the case-determining condition [23].
Observation period
The observation period for all subjects encompassed pre- and post-HCV diagnosis periods. For both cases and controls, the pre-diagnosis period was fixed at 365 days based on the study design, while the post-HCV diagnosis period was variable.
Outcomes
Treatment initiation, the primary outcome and grouping variable for this study, was assessed by the presence/absence of at least one prescription for IFN- alpha therapy during the post-HCV diagnosis period.
Independent variables
The independent variables were the factors that could be associated with the prescription/non-prescription of any of the IFN- alpha drugs used to treat HCV.
Demographic factors
Age, sex and race ethnicity were studied. Age, computed as of the date of the index HCV diagnosis, was categorized as: <=18, 19-44, 45-64 and >=65 years.
Substance abuse/dependence
Three types of substance abuse or dependence were examined based on primary or secondary ICD-9 codes in claims filed during the observation period: alcohol, opioid and other illicit drugs. Active substance or alcohol abuse can be considered a contraindication to IFN- alpha treatment [2].
Psychotropic drug use
The use of three types of psychotropic drugs was identified: antidepressants, antipsychotics and anxiolytics. Antidepressant use was of particular interest as a proxy for depression because the potential adverse effects of IFN- alpha on the course of depression have received considerable attention in research [7,24-26]. In addition, high rates of psychiatric disorder have been reported in studies of veterans with HCV [14,15].
Selected chronic comorbid conditions
Since the Charlson comorbidity index has been adapted for use in administrative claim databases, the component chronic conditions are linked to specific ICD-9 codes [27,28]. The relationship between each of the chronic comorbid conditions in the index and the outcome variables was examined in this study.
A number of chronic conditions may be triggered by or exacerbated by HCV treatment including: depression, any autoimmune disorder, ischaemic heart disease, conditions of compromised immune status, haemolytic anaemia and respiratory conditions. These conditions were assessed based on the relevant ICD-9 codes.
Medical utilization
The indicators of medical utilization were: the occurrence of hospitalization (yes/no) and the numbers of physician office visits, emergency visits and surgical procedures excluding liver biopsy. These general utilization dimensions may suggest increased access to levels of treatment (e.g. physician office care vs emergency care) or serve as proxies for health status (hospitalized patients may be more health-compromised than those not hospitalized). In addition, the occurrence of liver biopsy has particular relevance to the evaluation of liver disease and a consideration of the benefits of treatment with IFN [2].
Liver disease variables
Only chronic liver disease variables as defined in the Charlson comorbidity index were used in the analyses. Based on severity, there are two liver-related conditions in the index [27,28]. 'Mild' liver disease consists of ICD-9 codes for alcoholic cirrhosis, cirrhosis without mention of alcohol, biliary cirrhosis and chronic hepatitis, while 'moderate/severe' liver disease encompasses ICD-9 codes for hepatic coma, portal hypertension, other sequelae of chronic liver disease and oesophageal varices. These liver conditions could be linked to other conditions and not associated with HCV.
Statistical analysis
Rate of IFN treatment
To calculate the rate of IFN- alpha treatment, the denominator was the entire eligible cohort of patients while the numerator consisted of all cases (patients with one or more IFN- alpha claims).
Predictors of IFN treatment status
All independent variables were examined for possible associations with prescription/non-prescription of IFN. Accordingly, crosstabs with chi-square tests were performed for each independent variable by IFN status. Unadjusted logistic regression with the prescription/non-prescription of any of the HCV medications as the dependent variable was conducted for each independent variable using index referencing [29]. To calculate odds ratios (OR) with 95% confidence intervals (CI), each variable category was compared with subjects in the index category.
Analysis by domain limited the number of statistical tests while facilitating the identification of risk factors for treatment that were adjusted for demographic differences. Independent variables were divided into five domains that included: substance abuse or dependence, psychotropic drug use, chronic conditions in the Charlson comorbidity index, chronic conditions triggered by or exacerbated by IFN- alpha and medical utilization factors. Within each variable domain, adjusted (multivariate) analyses were conducted using logistic regression with IFN status as the dependent variable. Each model included all domain-specific variables, as well as all demographic variables.
In the summary multivariate phase, all variables that had significant effects in the domain-specific analyses were entered into a comprehensive logistic regression model that also included age, sex and race/ethnicity dimensions. Subsequently, a final summary model was generated through backwards stepwise elimination that included only significant variables related to IFN- alpha treatment status.
All tests of significance were conducted at an alpha level of 0.05 (two-tailed) with no correction to the alpha level based on multiple statistical comparisons. Confidence intervals that excluded unity were considered statistically significant at an alpha level of 5%. The subject, and not the claim, always served as the unit of analysis.
REFERENCES
1 Everson GT, Weinberg H. Living with Hepatitis C: A Survivor's Guide, 3rd edn. New York: Hatherleigh Press, 2002.
2 National Digestive Diseases Information Clearinghouse (NIDDK). Chronic Hepatitis C: Current Disease Management. NIH Pubication No. 03-4230. February 2003. http://www.niddk.nih.gov/health/digest/pubs/chrnhepc/chrnhepc.htm (accessed on 5/7/2003).
3 Seeff LB, Hoofnagle JH. Appendix: The National Institutes of Health Consensus Development Conference Management of Hepatitis C 2002. Clin Liver Dis 2003; 7: 261-287.
4 Fried MW, Shiffman ML, Reddy R et al. Peginterferon alfa-2a plus ribavirin for chronic hepatitis C virus infection. N Engl J Med 2002; 347: 975-982.
5 Manns MP, McHutchison JG, Gordon SC et al. Peginterferon alfa-2b plus ribavirin compared with interferon alfa-2b plus ribavirin for initial treatment of chronic hepatitis C: a randomized trial. Lancet 2001; 358: 958-965.
6 Roche Pharmaceuticals. Product information for PEGASYS©. Roche Pharmaceuticals. December, 2002. URL: http://www.rocheusa.com/products/pegasys/pi.pdf (accessed on 2/5/2003).
7 Bonaccorso S, Puzella A, Marino V et al. Immunotherapy with interferon-alpha in patients affected by chronic hepatitis C induces an intercorrelated stimulation of the cytokine network and an increase in depressive and anxiety symptoms. Psychiatry Res 2001; 105: 45-55.
8 Downs AM, Dunhill MG. Exacerbation of psoriasis by interferon-alpha therapy for hepatitis C. Clin Exp Dermatol 2000; 25: 351-352.
9 Georgetson MJ, Yarze JC, Lalos AT et al. Exacerbation of psoriasis due to interferon-alpha treatment of chronic active hepatitis. Am J Gastroenterol 1993; 88: 1756-1758.
10 Taylor C, Burns DA, Wiselka MJ. Extensive psoriasis induced by interferon alfa treatment for chronic hepatitis C. Postgrad Med J 2000; 76: 365-367.
11 Unoki H, Moriyama A, Tabaru A et al. Development of Sjogren's syndrome during treatment with recombinant human interferon-alpha-2b for chronic hepatitis C. J Gastroenterol 1996; 31: 723-727.
12 Passos de Souza E, Evangelista Segundo PT, Jose FF et al. Rheumatoid arthritis induced by alpha-interferon therapy. Clin Rheumatol 2001; 20: 297-299.
13 Pittau E, Bogliolo A, Tinti A et al. Development of arthritis and hypothyroidism during alpha-interferon therapy for chronic hepatitis C. Clin Exp Rheumatol 1997; 15: 415-419.
14 Muir AJ, Provenzale D. A descriptive evaluation of eligibility for therapy among veterans with chronic hepatitis C virus infection. J Clin Gastroenterol 2002; 34: 268-271.
15 Nguyen HA, Miller AI, Dieperink E et al. Spectrum of disease in US veteran patients with hepatitis C. Am J Gastroenterol 2002; 97: 1813-1820.
16 Nicklin DE, Schultz C, Brensinger CM et al. Current care of hepatitis C-positive patients by primary care physicians in an integrated delivery system. J Am Board Fam Pract 1999; 12: 427-500.
17 Rosenberg DM, Cook SF, Lanza LL. Health care, treatment patterns and cost of services for patients infected with chronic hepatitis C virus in a large insured New England population. J Viral Hepat 2000; 7: 361-367.
18 Friedman LM, Furberg CD, DeMets DL. Fundamentals of Clinical Trials. Littleton, MA: PSG Publishing Company, Inc., 1985.
19 Strader DB. Understudied populations with hepatitis C. Hepatology 2002; 36(5 Suppl 1): S226-S236.
20 ICD-9-CM. Professional for Hospitals, Vol. 1, 2 and 3. In: Hart AC, Hopkins AC (eds). International Classification of Diseases, Clinical Modification, 9th revision, 6th edn. Salt Lake City, UT: Ingenix/St. Anthony Publishing/Medicode, 2002; 1: 13.
21 Kelsey JL, Whitemore AS, Evans AS et al. Methods in Observational Epidemiology, 2nd edn. New York: Oxford University Press, 1996.
22 Szklo M, Nieto FJ. Epidemiology: Beyond the Basics. Gaithersburg, MD: Aspen Publishers, Inc, 2000.
23 Kleinbaum DG, Kupper LL, Morgentern H. Epidemiologic Research: Principles and Quantitative Methods. Belmont, CA: Lifetime Learning Publications, 1982.
24 Koskinas J, Merkouraki P, Manesis E et al. Assessment of depression in patients with chronic viral hepatitis: effect of interferon treatment. Dig Dis 2002: 20: 284-288.
25 Kraus MR, Schafer A, Faller H et al. Psychiatric symptoms in patients with chronic hepatitis C receiving interferon alfa-2b therapy. J Clin Psychiatry 2003; 64: 708-714.
26 Zdilar D, Franco-Bronson K, Buchler N et al. Hepatitis C, interferon alfa, and depression. Hepatology 2000; 31: 1207-1211.
27 Charlson ME, Pompei P, Ales KL et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis 1987; 40: 373-383.
28 Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM Administrative databases. J Clin Epidemiolol 1992; 45: 613-619.
29 Stokes ME, Davis CS, Koch GG. Categorical Data Analysis using the SAS System. Cary, NC: SAS Institute Inc., 1995.
30 National Institute of Health Consensus Statement. Management of Hepatitis C: 2002, Vol. 19, No 1, June 10-12, 2002. http://consensus.Nih.gov/cons/116/116cdc_intro.htm (accessed on 5/1/2003).
31 Capuron L, Hauser P, Hinze-Selch D, Miller AH, Neveu PJ. Treatment of cytokine-induced depression. Brain Behav Immun 2002; 16: 575-580.
32 Musselman DL, Lawson DH, Gumnick JF et al. Paroxetine for the prevention of depression induced by high-dose interferon alfa. N Engl J Med 2001; 344: 961-966.
33 Dieperink E, Willenbring M, Ho SB. Neuropsychiatric symptoms associated with hepatitis C and interferon alpha: a review. Am J Psychiatry 2000; 157: 867-876.
34 Hauser P, Khosla J Aurora H et al. A prospective study of the incidence and open-label treatment of interferon-induced major depressive disorder in patients with hepatitis C. Mol Psychiatry 2002; 7: 942-947.
35 Harris HE, Ramsay ME, Andrews N et al. Clinical course of hepatitis C virus during the first decade of infection: cohort study. Br Med J 2002; 324: 1-6.
36 Minola E, Prati D, Suter F et al. Age at infection affects the long-term outcome of transfusion-associated chronic hepatitis C. Blood 2002; 99: 4588-4591.
37 Hamada H, Yatsuhashi H, Yano K et al. Impact of aging on the development of hepatocellular carcinoma in patients with posttransfusion chronic hepatitis C. Cancer 2002; 95: 331-339.
38 Fleiss JL. Statistical Methods for Rates and Proportions, 2nd edn. New York: John Wiley & Sons, 1981.
39 Hoofnagle JH. Course and outcome of hepatitis C. Hepatology 2002; 36(5 Suppl 1): S21-S29.
40 Licata A, Di Bona D, Schepis F et al. When and how to treat acute hepatitis C? J Hepatol 2003; 39: 1056-1062.
41 Kirking DM, Ammann MA, Harrington CA. Comparison of medical records and prescription claims files in documenting prescription medication therapy. J Pharmacoepidemiol 1996; 5: 3-15.
42 West SL, Strom BL, Poole C. Validity of pharmacoepidemiology drug and diagnosis data. In: Strom BL, ed. Pharmacoepidemiology (third edition). New York: John Wiley & Sons, Ltd, 2000: 661-706.
|
|
| |
| |
|
|
|
