Updated Direct Costs of Medical Care for HIV-Infected Patients Within a Regional Population From 2006 to 2017

HB Krentz; Q Vu; MJ Gill


HIV Medicine. 2020;21(5):289-298. 

In This Article

Abstract and Introduction


Objectives: The aim of the study was to reappraise the precise costs of HIV care and cost drivers, to determine the optimal tools for modelling costs for HIV care, and to understand the implications of changing medical management of HIV-infected patients for both subsequent outcomes and health care budgets.

Methods: We obtained all drug, laboratory, out-patient and in-patient care costs for all HIV-infected patients followed between 1 January 2006 and 31 December 2017 (2017 Cdn$). Mean cost per patient per month (PPPM) was used as the standard comparator value. Patients were stratified based on CD4 count: (1) ≤ 75, (2) 76–200, (3) 201–500 and (4) > 500 cells/μL. We determined the cost for only HIV-related expenses. We compared current costs with costs previously reported for the same population.

Results: The number of HIV-infected patients in care doubled from 2006 to 2017; total costs increased from $12.4 to $30.1 million, with antiretroviral (ARV) drugs accounting for 78.8% of costs by 2017. Out-patient/laboratory costs declined from 12% to 8.5%, while in-patient costs exhibited more annual variation. Mean PPPM costs increased from $1316 in 2006 to $1712 in 2014, declining to $1446 in 2017. Higher PPPM costs were associated with CD4 counts < 200 cells/μL. Costs have shifted. While the cost of ARV drugs increased by 32%, the costs of out-patient and in-patient services decreased by 80% and 71%, respectively. Most of the decrease for in-patient costs was attributable to a substantial decrease in HIV-related hospitalizations.

Conclusions: Although antiretroviral therapy (ART) provides immense benefits, it is not inexpensive. ARV drugs remain the largest cost driver. Hospital costs have remained low. Substantial costs of lifelong ART necessitate innovative, locally applicable strategies for ARV selection and use.


Over the last decade, medical management of HIV infection has changed dramatically.[1–6] Antiretroviral therapy (ART) is now routinely offered, regardless of the CD4 count, soon after HIV diagnosis.[7] Lifelong, continuous ART is then recommended.[8–10] New classes of antiretroviral (ARV) drugs, with improved potency, dosing schedules, and tolerability but decreased toxicity and susceptibility to drug–drug interactions, often conveniently co-formulated, have significantly decreased morbidity and mortality, making the goal of lifelong therapy achievable for many persons living with HIV (PLWHIV).[11] Recent studies suggest that this care approach will, for many PLWHIV, preserve health and restore life expectancy back to close to normal, while also reducing the risk to public health from untreated infection.[12,13] These benefits have now led to such improved longevity that new initiatives are being promoted to care for the aging HIV-infected population.[14–17]

The full implications for health budgets of these dramatic advances and shifts in care have so far been inadequately explored but need to be monitored closely as a consequence of the substantial current and predicted future cost of HIV care. Many new approaches and considerations to manage costs have recently entered the cost landscape, such as the use of new, less expensive ARVs, dual versus triple therapy regimens, and generic formulations of older ARVs[18–21] as well as a proposed reduced frequency of HIV care visits and testing.[22,23] Earlier approaches for predicting costs based on CD4 counts[24,25] have become problematic as few patients experience true virological failure. Even the ability to capture costs precisely may stress current costing data sets as a consequence of the new need to capture details of generic or innovative manufactured ARVs, the use of co-formulated or de-simplified treatments, and complex on/off drug regimens[26] and to accommodate the fluid costing of generic ARVs.[27] Attribution of visits and laboratory testing to HIV or comorbidity care adds even more complexity to the precise costing of HIV care in the current era.

The labile nature of these costs led us to examine all costs of HIV care up to December 2017. Previously, we had speculated that the movement to greater testing and earlier initiation of ART could, despite the costs of increased ART use, be partly counterbalanced by reduced total medical costs from the enhanced health of those on ART with a diminishing number of presentations with AIDS.[28] We wished to examine these hypotheses and to measure shifts, in the current era, of the cost components of HIV care. For comparative purposes, we followed our earlier costing model that had looked at all medical services provided to PLWHIV living in southern Alberta, Canada, between 1997 and 2006.[28] By using the same approach in the same population, we could evaluate the costs of HIV care from the start of ART (i.e. 1997) to the present. By examining costs divided into costing category, clinical category (i.e. CD4 count), and demographic category, we wished to understand the implications of these trends for costs and outcomes over time. This measured approach, using direct rather than estimated costs of medical care, allows for a better and perhaps more precise approach to the dynamic nature of the medical management of HIV infection and health care budgets.

Methods. The direct cost of care was obtained for all PLWHIV > 15 years old followed at Southern Alberta Clinic (SAC) in Calgary, Alberta, Canada between 1 January 2006 and 1 January 2018. All eligible (i.e. Alberta residents) PLWHIV within the region receive HIV care through SAC at no personal cost under universal health care, including the cost of all ARV drugs, out-patient clinic visits, laboratory tests, and in-patient hospital care. Patients were included in the study if they had at least one regular clinic visit within a calendar year and were followed until they died, moved, or became lost to follow-up (i.e. no regular clinic visits for 12 months). Sociodemographic and clinical data are directly updated at every clinic visits in SAC's administrative database. Sociodemographic data included gender, age (as of 31 December), self-reported ethnicity, and most likely mode of HIV transmission. Health care utilization (per patient per year) included the number of HIV care visits, the number of CD4 counts, viral load measurements and other laboratory tests performed, and the number of days of HIV- and non-HIV-related in-patient hospital admission. Clinical data included CD4 count, viral load status, AIDS diagnosis, ARV use (naïve/experienced), and number of ARV drugs per regimen as of 31 December of each respective year.

Costing Categories. To facilitate costing comparisons with our historical data,[28] we again obtained the direct costs of medical care from original costing sources. Drug costs, laboratory test utilization, and out-patient care costs were derived directly from the SAC pharmacy, Calgary Laboratory Services, and the SAC costing database, whereas in-patient costs (i.e. unit service costs) were supplied by Alberta Health Services (AHS). The unit service costs used are market values charged to the regional payer (AHS). All costs were obtained directly from the costing agencies. In order to more accurately compare costs over the 12 years, we converted all costs to 2017 Canadian dollars adjusting for inflation.

ARV and non-ARV drug costs were based on 1 month's supply of medication. Non-ARV drug costs were obtained from the central database based on wholesale values and average commercial mark-ups, whereas the costs of ARV drugs were obtained directly from the SAC pharmacy database. Costs associated with out-patient care included the costs of all regular HIV clinic visits, visits to HIV-related specialists, and all laboratory testing. Laboratory test costs included the costs of CD4 count, viral load measurement, HIV genotypic resistance testing, serological tests, haematology and routine chemistry testing. Costs were calculated as cost per test per patient. All in-patient care costs regardless of admitting diagnosis were obtained from the Data Integration Measurement and Reporting (DIMR) department of Alberta Health Services. These costs were calculated on a 'per specific patient per day' basis based on the length of the hospital admission and included both the direct costs associated with patient care (i.e. costs of drugs, supplies, equipment, salaries, nursing care, and laboratory and diagnostic testing) and indirect costs associated with hospital overheads (i.e. costs of administration and support services). In-patient visits were categorized as HIV related if the most responsible diagnosis was AIDS-related or a condition directly attributable to HIV disease. To confirm correct coding of diagnosis, we also reviewed the hospital discharge summaries of all SAC patients to verify whether they were HIV- or non-HIV-related admissions.

We used the mean cost per patient per month (PPPM) of medical care as the standard comparator value. PPPM cost was determined by multiplying service usage by unit service cost divided by the total number of SAC patients followed during the year. Patients were stratified into four CD4 count categories based on their status at the end of each month: (1) ≤ 75, (2) 76 to 200, (3) 201–500 and (4) > 500 cells/μL. Previously, we had subdivided patients into two treatment groups: (1) ARV-naïve (i.e. never on ARV) and (2) ARV-experienced (i.e. either currently or previously on ARVs). As a consequence of the small number and proportion of ARV-naïve patients, we did not subdivide the recent population into these categories.

To separate costs attributable only to HIV infection or related comorbidities, we reanalysed the data removing all presumed non-HIV-related costs, including the cost of non-ARV drugs, non-HIV physician visits and non-HIV-related hospitalizations. The costs of ARV drugs, HIV-related out-patient visits, and HIV-related in-patient hospitalizations were considered 'HIV related'.

We performed a historical comparison of costs associated with HIV infection between our current study and our previous study[28] that examined costs within the same regional population and using the same methodology. Although the study covers 1997–2006, we compared only years 2005 and 2017 adjusted to 2017 Cdn$.

Basic descriptive statistics [i.e. mean, standard error (SE) and median] for all demographic and clinical characteristics were used. Where appropriate, we used Student t-tests for normally distributed data and Mann–Whitney U-tests for non-normally distributed variables. Statistical significance was defined as P < 0.05 and we used 95% confidence intervals. Statistical analyses were performed with SPSS v20.0 (IBM, Armonk, NY).

Patients attending SAC voluntarily provide informed consent authorizing the use of confidential administrative data for research purposes. The use of nonnominal demographic and clinical data has been approved for research purposes by the University of Calgary Conjoint Heath Research Ethics Board.