Despite advances in immune reconstitution from highly active antiretroviral treatment (HAART), which was first introduced in 1996, 1,2 cytomegalovirus (CMV) retinitis remains the leading cause of blindness among patients with AIDS. New cases of CMV retinitis continue to be identified due to late diagnosis of HIV and or AIDS, poor adherence to HAART treatment, and viral resistance to one or more components of HAART treatment. 3 With the advent of new Food and Drug Administration–approved anti-CMV therapies, such as oral valganciclovir, 4 ganciclovir intraocular implant, 5 fomivirsen intraocular injection, 6 and intravenous cidofovir, 7 providers and patients with CMV retinitis have more treatment options available today than in previous years of the HIV/AIDS epidemic. Along with disease- and patient-related factors, costs of treatment may be an important consideration in choosing the type of medical therapy.
Prior economic analyses of treatment of CMV retinitis have shown that oral treatment is associated with cost savings as compared with intravenous treatment. 8–14 These studies noted that the higher incidence of adverse events among the intravenous treatment group had increased the costs of treating CMV retinitis compared with oral therapy. However, these studies were performed prior to the advent of newer anti-CMV therapies and most were done before HAART became well established. For example, orally administered valganciclovir appeared to be equivalent to intravenous ganciclovir for induction therapy and is a convenient therapy for CMV retinitis patients with AIDS. 4 The use and economic impact of these newer agents are unclear.
We evaluated the epidemiology of CMV retinitis, treatment patterns, and the costs of CMV retinitis among a national managed-care population in the post-HAART era. Because mode of administration has important implications in clinical management of CMV retinitis, we also compared the cost consequences of oral, intravenous, and intraocular treatments.
Study Population and Design
We conducted a retrospective cohort analysis using a large, national claims database of commercially insured enrollees (Pharmetrics, Inc., Boston, MA). This database contained claims data from >32 million plan members enrolled in 57 health plans across the United States. Data were collected from the years 1997 to June 2002.
We included members who had claims with diagnosis codes for HIV/AIDS (any International Classification of Disease [ICD]-9 code starting with 042, v08, or 079.53), CMV (ICD-9 code 078.5), retinitis (any ICD-9 code starting with 362 or 363), and had received anti-CMV treatment. Anti-CMV treatment was identified by national drug code or the Health Care Financing Administration’s Common Procedure Coding System (HCPCS) for intravenous and oral ganciclovir (Cytovene, Roche Laboratories, Inc., Nutley, NJ), oral valganciclovir (Valcyte, Roche Laboratories, Inc., Nutley, NJ), intravenous foscarnet (Foscavir, Astra Zeneca International, Inc., Wilmington, DE), intravenous cidofovir (Vistide, Gilead, Forest City, CA ), ganciclovir implant (Vitrasert, Roche Laboratories, Inc., Nutley, NJ), or fomivirsen treatment (Vitravene, ISIS Pharmaceuticals, Inc., Carlsbad, CA). Because there were occasional nonspecific claims for intravenous treatment with a diagnosis code of CMV or retinitis during a CMV retinitis episode of care, we included these claims as “unspecified intravenous anti-CMV treatment.”
Early symptoms without later confirmation of diagnosis by an ophthalmologist may lead to empiric anti-CMV therapy. Therefore, we took measures to ensure that only patients with established CMV disease were included. We did so by excluding cases that did not have at least 1 repeat anti-CMV treatment dose, were censored within 30 days of the incident anti-CMV treatment, and whose incident treatment did not fall within 30 days of a retinitis diagnosis.
Every case that satisfied the inclusion and exclusion criteria was defined a CMV retinitis episode of care. Based on expert opinion (K.G. and J.D.), we defined an episode of care as starting 21 days prior to the date of incident anti-CMV treatment and lasting until 30 days past the last contiguous dose of oral or intravenous treatment or 270 days after the last intraocular treatment, whichever came later. As prescriptions may have delays, we allowed for a 21-day grace period between prescriptions to prevent premature termination of the episode of care. The criteria for an incident anti-CMV treatment claim were no oral or intravenous treatment of CMV in the previous 51 days (1 month + grace period) and no intraocular treatment of CMV in the prior 291 days (9 months + grace period). Using this algorithm, each patient could have ≥1 CMV retinitis episode of care during the analysis period.
We calculated the prevalence and incidence of HIV/AIDS in the national managed-care database as well as the prevalence and incidence of CMV retinitis among those with HIV/AIDS. Prevalence of HIV/AIDS was defined as the number of patients with a diagnosis of HIV or AIDS over the total number of patients in the database for a given year. Incident cases of HIV/AIDS were similarly defined and required no prior HIV or AIDS diagnoses codes in at least the preceding 12 months or more. Prevalence of CMV retinitis was defined as the number of patients satisfying our inclusion criteria of CMV retinitis who received care over the total number of patients with HIV/AIDS for a given year. Incident cases of CMV retinitis were similarly defined and required an absence of CMV or retinitis diagnoses in at least the preceding 6 months.
After identifying members who had a CMV retinitis episode of care, we calculated the average cost per patient, cost per episode, and cost per day. Costs were further stratified into costs of drug treatment, infusion (intravenous only), adverse events, phlebitis (intravenous only), and outpatient and inpatient care.
To evaluate the cost consequences of oral vs. intraocular vs. intravenous treatment, we accounted for patients who received ≥1 treatment modality by analyzing treatment type in a time-dependent format. For example, a person who started on oral treatment and then switched to intraocular treatment would contribute periods to each of these treatment types. All CMV retinitis episodes of care were categorized into either oral treatment time, intravenous treatment time, intraocular treatment time, or combinational treatment time (defined as the administration of ≥2 of the aforementioned therapies simultaneously). If a switch occurred between treatment types, the 2-week period after the switch was considered a washout period. Events in the washout period were attributed to the preceding treatment.
To identify costs of adverse events, we searched for ICD-9 codes for the 5 most frequently reported adverse events in each product’s package insert. Only claims with these ICD-9 codes that started during a period corresponding to exposure to a potential offending anti-CMV product were considered as an adverse event. All claims with a similar ICD-9 code within 14 days after the first code were included in the cost of the adverse event.
Because opportunistic infections may increase health care costs, in multivariate regression analyses we adjusted our analyses for co-existing AIDS-defining illnesses using the 1993 Centers for Disease Control revised classification system definitions. 15
All costs were adjusted to 2002 US dollars using the medical component of the consumer price index. We employed a nonparametric bootstrapping analysis to test for significance among treatment groups. The nonparametric approach does not make assumptions about the normality of the cost data or its variance. 16 The bootstrap estimates of the regression and the tests of significance were prepared with the use of the “bootcov()” function in the HMISC library of SPLUS software (Insightful Corp., Seattle, WA). We adjusted for age, sex, geographic region, primary insurance provider, proportion of time on HAART treatment, frequency of coexisting AIDS defining illness, and treatment duration in our multiple linear regression models.
Among 32 million unique members in the national claims database, a total of 27,350 had a diagnosis claim of HIV or AIDS. Among this group, 592 had at least 1 claim with a diagnosis of CMV. There were 135 patients who met the additional inclusion criteria for retinitis diagnoses and anti-CMV treatment. Of these cases, we excluded 37 patients because 19 did not have repeat anti-CMV treatment, 12 did not have treatment coinciding with retinitis diagnoses, and 6 were censored 30 days after initial retinitis claim. The remaining 98 patients formed the cohort for the analysis (Table 1).
While the prevalence and incidence of enrolled HIV or AIDS cases increased, the corresponding rates for CMV retinitis decreased during the study period (Fig. 1). The prevalence of enrolled HIV or AIDS members in the managed-care population increased from 524 per million in 1997 to 927 per million in 2001. The incidence of newly enrolled patients with HIV or AIDS increased from 7 enrolled members per million in 1997 to 150 enrolled members per million in 2001. The prevalence rate of CMV retinitis decreased from 214 per 10,000 HIV or AIDS members in 1997 to 41 per 10,000 HIV or AIDS members in 2001. Similarly, the incidence of CMV retinitis decreased from 23 per 10,000 HIV or AIDS members in 1997 to 8 per 1000 HIV or AIDS members in 2001.
Figure 2 shows the percentage of all drug claims for oral, intravenous, and intraocular treatment and the proportion by specific product type. The proportion of claims that were for oral anti-CMV treatment increased from 9.6% in 1997 to 43.4% in 2002 while the proportion of intravenous treatment claims decreased from 77.1% in 1997 to 40.8% in 2002. During this period, the proportion of intraocular treatment claims remained relatively constant, increasing modestly from 13.3 to 15.8%.
The average cost per patient with CMV retinitis was $23,172 (SD of $29,520) and the median cost was $14,870. The average cost per episode of care was $19,576 (SD of $24,846) and the median cost was $12,574. The total cost per day was approximately $105 dollars. Table 2 details the total costs incurred stratified by cost category, the duration in days of therapy, and the cost per day for each treatment type. The costs per day were $69 for oral therapy, $163 for intravenous, and $70 for intraocular.
We adjusted for differences in the proportion of patient-time on HAART treatment, proportion of time with coexisting AIDS-defining illness, duration of treatment periods, and other demographics (Table 3). Oral treatment saved $7185 per treatment period as compared with intravenous treatment (95% CI, -$11,342, -$3029). Intravenous treatment costs $6,866 (95% CI, $2161, $11,570) more than intraocular treatment. The total cost differences between oral and intraocular treatment groups were not statistically significant. Because combinational treatment represented a small fraction of all treatment, we do not present cost differences between this group and the others.
In this study, we found that total costs of treatment were lower for oral and intraocular therapy as compared with intravenous therapy. After adjustment for potential demographic and treatment confounders, the costs for intravenous treatment still remained higher. The magnitude of the difference between intravenous treatment and the other groups was large. The cost differences were mainly driven by higher outpatient costs in the intravenous treatment group. The higher costs for phlebitis for intravenous treatment were also statistically significant compared with the other groups.
It has been demonstrated that intravenous ganciclovir and foscarnet are comparable in their efficacy. 17 The efficacy of oral valganciclovir, which was approved for the treatment of CMV retinitis in 2001, is also comparable to that of intravenous ganciclovir for induction therapy. 4 Nevertheless, our data show that the proportion of cases treated with intravenous treatment remained relatively high in 2002 (Fig. 2), at approximately 40%. Switching these patients to oral or intraocular therapy represents an opportunity for cost savings.
If “local” therapy (intravitreous fomivirsen or ganciclovir implant) is chosen, concurrent oral ganciclovir is generally recommended to reduce the risk of second-eye or visceral CMV disease, and systemic ganciclovir also significantly reduces the risk of developing Kaposi sarcoma. 18 While this approach may be more effective for the general health of a patient than local therapy alone, it also adds to the cost of therapy and may expose the patient to more treatment-related adverse events. Because too few patients were on simultaneous treatment, the study was not powered to examine costs differences among various dual-treatment combinations.
Prior to HAART therapy, Palella et al 2 demonstrated a decline in opportunistic infections among the HIV/AIDS population. Our study showed further, rather dramatic, decreases in CMV retinitis in a managed-care HIV population during the post-HAART era. Our projected rate of CMV retinitis decline from 1997 to 1998 may be artificially accentuated. We attribute this to a smaller number of managed-care plans that supplied data in 1997 in addition to the low number of enrolled HIV or AIDS patients in a plan. However, after 1997, the incidence rates associated with CMV retinitis in our analysis are consistent with HIV Outpatient Study (HOPS) 19 database, which demonstrated an approximate incidence of 50–300 CMV retinitis cases per 10,000 AIDS patients from 1998–2001.
Our study had several limitations. First, because our data source was claims based, we do not know the specifics of the underlying clinical indicators that could explain cost differences among the treatment groups. For example, intravenous treatment may not be the cause of higher costs but a reflection of worsening disease states, which has higher costs. However, we did adjust for various proxy confounders such as coexisting AIDS-defining illnesses and use of HAART treatment, which did not change the large effect size seen in our analysis. Secondly, we used strict inclusion criteria, which may have limited our sample and the generalizability of our results. Finally, we cannot be certain that miscoding did not occur in some of these claims. However, use of claims codes in HIV was found to be very accurate in prior analyses. 20,21
In conclusion, we observed that CMV retinitis is less prevalent and incident among a national managed-care population during the HAART era of the HIV/AIDS epidemic. The total cost per episode of care, however, remains high. We found that oral and intraocular treatment modalities led to lower total costs, suggesting that these agents maybe preferable as initial treatment over intravenous therapy.
The authors thank Ms. Kathy Fraeman and Dr. Chris Barker for their assistance with statistical programming and analyses.
1. Cunningham WE, Markson LE, Andersen RM, et al. Prevalence and predictors of highly active antiretroviral therapy
use in patients with HIV
infection in the united states. HCSUS Consortium. HIV
Cost and Services Utilization. J Acquir Immune Defic Syndr
2. Palella FJ Jr, Delaney KM, Moorman AC, et al. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. HIV
Outpatient Study Investigators. N Engl J Med
. 1998;338: 853–860.
3. Dunn JP. Cytomegalovirus retinitis
in 2003. Hopkins HIV Rep
. 2003;15: 8–11.
4. Martin DF, Sierra-Madero J, Walmsley S, et al. A controlled trial of valganciclovir as induction therapy for cytomegalovirus retinitis
. N Engl J Med
5. Martin DF, Dunn JP, Davis JL, et al. Use of the ganciclovir implant for the treatment of cytomegalovirus retinitis
in the era of potent antiretroviral therapy: recommendations of the International AIDS
Society-USA panel. Am J Ophthalmol
6. Piascik P. Fomiversen sodium approved to treat CMV retinitis. J Am Pharm Assoc
7. MacDougall DS. Update on cidofovir for CMV retinitis. J Int Assoc Physicians AIDS Care
8. Davies L, Maynard A. An economic exploration of oral and intravenous ganciclovir in the induction and maintenance treatment of AIDS
-related cytomegalovirus retinitis
. Int J STD AIDS
9. Rachlis A, Smaill F, Walker V, et al. Incremental cost-effectiveness analysis of intravenous ganciclovir versus oral ganciclovir in the maintenance treatment of newly diagnosed cytomegalovirus retinitis
in patients with AIDS
10. Lee TA, Sullivan SD, Veenstra DL, et al. Economic evaluation of systemic treatments for cytomegalovirus retinitis
in patients with AIDS
11. Sullivan SD, Mozaffari E, Johnson ES, et al. An economic evaluation of oral compared with intravenous ganciclovir for maintenance treatment of newly diagnosed cytomegalovirus retinitis
patients. Clin Ther
12. Griffiths RI, Bleecker GC, Jabs DA, et al. Pharmacoeconomic analysis of 3 treatment strategies for cytomegalovirus retinitis
in patients with AIDS
13. Hardens M. Ganciclovir evaluation in AIDS
patients with cytomegalovirus retinitis
: a European study of treatment patterns and resource utilization. AIDS
. 1996;10(suppl 4):S25–S30.
14. Graf von den Schulenburg JM, Wahling S, Stoll M. German health economic cost evaluation on oral ganciclovir in treating cytomegalovirus retinitis
15. Centers for Disease Control and Prevention. 1993 Revised classification system for HIV
infection and expanded surveillance case definition for AIDS
among adolescents and adults. JAMA
16. Desgagne A, Castilloux AM, Angers JF, et al. The use of the bootstrap statistical method for the pharmacoeconomic cost analysis of skewed data. Pharmacoeconomics
17. Foscarnet-Ganciclovir Cytomegalovirus Retinitis
Trial. 4. Visual outcomes. Studies of Ocular Complications of AIDS
Research Group in collaboration with the AIDS
Clinical Trials Group. Ophthalmology
. 1994; 101:1250–1261.
18. Martin DF, Kuppermann BD, Wolitz RA, et al. Oral ganciclovir for patients with cytomegalovirus retinitis
treated with a ganciclovir implant. Roche Ganciclovir Study Group. N Engl J Med
19. Centers for Disease Control. HIV Outpatient Study Database
. Atlanta: Centers for Disease Control; 2003.
20. Fasciano NJ, Cherlow AL, Turner BJ, et al. Profile of Medicare beneficiaries with AIDS
: application of an AIDS
casefinding algorithm. Health Care Financ Rev
21. Rosenblum L, Buehler JW, Morgan MW, et al. HIV
infection in hospitalized patients and Medicaid enrollees: the accuracy of medical record coding. Am J Public Health