Abstract
We undertook a retrospective cohort study of patients with a positive HIV test in the emergency department who were then linked to care. Inpatient, outpatient, and emergency costs were collected for the first 2 years after HIV diagnosis. Fifty-six patients met the inclusion criteria; they were predominantly uninsured (73%) and African American (89%). The median total cost for a newly diagnosed patient over the first 2 years was US$36 808, driven predominantly by outpatient costs of US$17 512. Median inpatient and total costs were significantly different between the lowest (<200 cells/mm3) and highest (>499 cells/mm3) CD4 count categories (US$21 878 vs US$6607, P <.05; US$61 378 vs US$18 837, P <.05, respectively). Total costs were significantly different between viral load categories <100 000 HIV-RNA copies/mL and ≥100 000 HIV-RNA copies/mL (US$28 219 vs US$49 482, P <.05). Costs were significantly lower among patients diagnosed earlier in their disease. Decreased cost is another factor supporting early diagnosis and linkage to care for patients with HIV.
Introduction
In September 2006, the Centers for Disease Control and Prevention (CDC) recommended that HIV screening be a part of routine medical services and offered to every patient on a voluntary opt-out basis in order to identify patients early in the course of their disease. 1 Identifying patients early improves patient prognosis by linking them to care and starting antiretroviral therapy (ART). It also decreases transmission of the virus by patients through treatment 2 and awareness of infection. 3 However, early detection can increase the number of HIV-positive patients seeking care and can result in increased costs. Mathematical models 4 and other estimates using average visit and hospitalization costs 5,6 have suggested that an inverse relationship exists between annual costs and CD4 counts. Diagnosis at higher CD4 count has been described as an early presenter (CD4 count ≥500 cells/mm3) compared to a late presenter (CD4 count < 200 cells/mm3). The purpose of this study was to determine whether early detection of HIV can result in actual cost savings in a public charity health system with limited resources.
Methods
A retrospective cohort study was conducted using the medical records of patients who had a positive rapid HIV test in the emergency department (ED) of Interim Louisiana State University Public Hospital in the initial year of institution of the rapid test (2008). Fifty-six patients met the inclusion criteria of new diagnosis of HIV by rapid testing and linkage to care at the affiliated HIV clinic. The outpatient clinic affiliated with the hospital is the HIV outpatient program and is the largest single provider of HIV services in the region. Inpatient, outpatient, and emergency costs as well as number of visits were collected for each patient for the first 2 years after initial HIV diagnosis. Actual incurred costs were obtained for each patient from the financial department. These included costs of intensive care, inpatient and clinic visits, diagnostic testing including laboratory and imaging, as well as ancillary services such as mental health services and dental visits. Medications during inpatient stay were included in the costs analysis. Outpatient prescription medications were not included. SAS (SAS Institute Inc) was used for analysis and data were analyzed using the Kruskal-Wallis (KW) test to determine if there were differences in costs among the CD4 count categories (<200, 201-349, 350-499, >499 cells/mm3) at initial diagnosis. Significant KW results were supplemented by assessing pairwise differences with a Bonferroni correction to adjust for multiple comparisons. Differences in cost among the 2 viral load categories (<100 000 HIV-RNA copies/mL and ≤100 000 HIV-RNA copies/mL) were assessed using the Wilcoxon rank sum test.
Results
Fifty-six patients met the inclusion criteria. The mean age was 41 with a range of 19 to 61 years. The patients were predominantly male (54%), African American (89%), and uninsured (27% had any insurance including Medicaid). The mean CD4 count was 265 cells/mm3 with a 95% confidence interval of 204 to 325 cells/mm3. In all, 43% met CD4 criteria for AIDS. The demographics by CD4 count category are displayed in Table 1. Patients with lower CD4 counts were more likely to be male and of older age.
Demographic Factors by CD4 Count Category and Viral Load Category.
Abbreviations: SD, standard deviation; STI, sexually transmitted infection.
The median inpatient cost per patient over 2 years was US$36 808 (range US$2685-US$305 380) divided into median costs for inpatient, US$8763 (range 0-US$252 167); outpatient, US$17 512 (range US$61-US$72 291); and ED US$2212 (range 0-US21 453). The median amount recovered over 2 years was US$1694 (range 0-US$164 250).
Over 2 years, the median number of inpatient days was 2 (range 0-65 days), outpatient visits 19 (0-166), and emergency department visits 2 (range 0-20). The details of costs by location and CD4 count category are provided in Table 2. The difference between the total costs across each distinct category was statistically significant (P value <.01). The difference between the highest and lowest CD4 count categories for inpatient costs was also statistically significant (P value = .01).
Median Inpatient Cost per Patient over 2 Years by CD4 Count and Viral Load Category.
Two viral load categories were assessed: <100 000 HIV-RNA copies/mL, which included 35 patients and >100 000 HIV-RNA copies/mL, which included 20 patients. The difference between the total median costs for viral load <100 000 HIV-RNA copies/mL and >100 000 HIV-RNA copies/mL was statistically significant with a P value of .03. The difference in inpatient costs between viral load categories trended toward significance (P value = .07). Table 2 displays the details of cost by viral load category.
Discussion
Early diagnosis and linkage to care results in decreased transmission 2 and decreased mortality due to early ART initiation and opportunistic infection prophylaxis. In this study, costs involved were significantly lower among patients diagnosed early in their disease. Patients with lower CD4 counts (<200 cells/mm3) had 3 times the median costs as those with higher CD4 counts (>499 cells/mm3) in the first 2 years, driven predominantly by inpatient costs. Patient with higher viral loads (another marker of disease progression) had almost twice the costs as those patients with lower viral loads.
Late presentation into medical care has been shown to increase health care utilization and expenditure. Previous studies demonstrated a range of increased costs from 1.5 to 4 times higher mean expenditures when comparing early and late presentation in the year following HIV diagnosis. 6,7
Our study demonstrates that the inverse relationship between total costs and CD4 count category exists on a continuum. Earlier diagnosis led to decreased costs between each of our specified CD4 count category. The results demonstrate the importance of diagnosis at the higher CD4 counts because early presentation significantly saved money over intermediate or late presentation.
A previous study has also reported an association of higher costs of care with an HIV-RNA of greater than 100 000 copies/mL. 5 The plasma viral load is an independent predictor of the development of an AIDS-defining illness and death. 8 A high viral load is suggested to be a marker of later presentation outside of acute infection. Therefore, patients with higher viral loads could be expected to have higher costs, as demonstrated in our study.
Our study had the following limitations. There were elements of resource utilization that we did not include such as home-based services and outpatient pharmacy utilization. Medication costs, especially antiretroviral (ARV) drugs, are a significant component of the overall HIV costs and have been estimated to be as high as 40% to 50%. 9 The ARV drugs were provided by multiple sources within our study including AIDS Drug Assistance Program, Medicaid, and other insurances. The costs of outpatient medication including ARV drugs were excluded from our study, as it focused on the cost of HIV care to the Louisiana Charity health system. The cost differentials by CD4 count were the actual hospital costs. For hospital admission and outpatient visits, all services provided by the Charity hospital system, including ancillary services such as mental health visits, nutritional counseling, and dentist visits were included. It is possible that patients were provided services at nonaffiliated clinics or medical centers, but this is traditionally uncommon. The Charity health system in New Orleans provides comprehensive care to predominantly uninsured patients and these patients most often receive care through this system or its affiliates. Patients could have also moved out of the area and these costs would not have been captured in our study.
Our study included all costs directly incurred by the Charity health system. Costs were not simulated or estimated. All patients in this study were diagnosed in the ED through rapid point-of-care testing. In following the CDC’s recommendation for routine screening for HIV infection in all patients aged 13 to 64 years, the cost of increased screening was incurred by the Charity health system. The health benefits of earlier diagnosis have been documented, 10 but, in our health system’s cost-conscious paradigm, it is also important to demonstrate cost savings in early diagnosis. Computer simulation models have projected the cost-effectiveness of routine HIV screening. For example, Paltiel et al reported cost-effectiveness ratios, in 2001 US dollars, for routine HIV testing among outpatients of US$36 000 per quality-adjusted life year (QALY) gained for a 3% prevalence of undiagnosed HIV infection and US$38 000 per QALY for a 1% prevalence. 11 These estimates were not specific to the ED and larger-scale data collection to assess the cost-effectiveness of ED-based screening efforts are needed. 12 The prevalence of HIV in the Charity ED for the year 2008 was 1.7%, with 1.3% newly diagnosed positives (8204 tested). 13 Though our sample size was smaller than previous studies also demonstrating the impact of disease stage on medical costs, our findings were illustrative to the Charity health system of the economic importance for HIV screening and earlier diagnosis. 6,14 Inpatient costs were significantly different between CD4 count of less than 200 cells/mm3 and CD4 count of greater than or equal to 500 cells/mm3. Outpatient costs, though not significant, decreased between each CD4 count category. Our results over a 2-year time period (total costs: CD4 count <200 cells/mm3 US$61 378, CD4 count 201-349 cells/mm3 US$37 485, CD4 count 350-499 cells/mm3 US$26 478, and CD4 count greater than or equal to 500 cells/mm3 US$18 837) are very similar to total costs in previous studies. 14 Patients in New Orleans who utilize care through the Charity health system are predominantly uninsured and the cost savings of early entry into medical care will be of direct financial benefit for the health system.
Health care expenditures are ever rising. Enthusiasm to actively diagnose and enroll new HIV-positive patients in clinical services can wane when resources are limited. This is especially true in Louisiana, which has the highest statewide rate of new HIV infections. 15 This study demonstrated a significant cost savings (as much as 3 times over a 2-year period) by diagnosing and linking patients early in their disease. Decreased cost is another factor supporting early diagnosis and linkage to care for patients with HIV.
Footnotes
Authors’ Note
LER and JH contributed to conceptualization and design of the project, summarization of data, and drafting of the manuscript. PSS and NVS contributed to supervision of project as well as critical review and revision of the manuscript. I. Pathmanathan and M. Katz contributed to data collection and feedback on design of project as well as critical review and revision of the manuscript. L. Myers completed statistical analysis and wrote biostatistical methods as well as contributed to the approval of final manuscript.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: supported in part by 1 U54 GM104940 from the National Institute of General Medical Sciences of the National Institutes of Health which funds the Louisiana Clinical and Translational Science Center.