HIV Treatment as Prevention

Expansion of HAART Coverage Is Associated with Sustained Decreases in HIV/AIDS Morbidity, Mortality and HIV Transmission: The ‘‘HIV Treatment as Prevention’’ Experience in a Canadian Setting Julio S.G. Montaner1,2*, Viviane D. Lima1,2, P. Richard Harrigan1,2, Lillian Lourenço1, Benita Yip1,

Bohdan Nosyk1,3, Evan Wood1,2, Thomas Kerr1,2, Kate Shannon1,2, David Moore1,2, Robert S. Hogg1,3,

Rolando Barrios1,5, Mark Gilbert4, Mel Krajden4, Reka Gustafson5, Patricia Daly5, Perry Kendall6

1 BC Centre for Excellence in HIV/AIDS, Providence Health Care, Vancouver, British Columbia, Canada, 2 Division of AIDS, Department of Medicine, University of British

Columbia, Vancouver, British Columbia, Canada, 3 Faculty of Health Sciences, Simon Fraser University, Vancouver, British Columbia, Canada, 4 BC Centre for Disease

Control, Vancouver, British Columbia, Canada, 5 Vancouver Coastal Health Authority, Vancouver, British Columbia, Canada, 6 Ministry of Health, Province of British

Columbia, Victoria, British Columbia, Canada

Abstract

Background: There has been renewed call for the global expansion of highly active antiretroviral therapy (HAART) under the framework of HIV treatment as prevention (TasP). However, population-level sustainability of this strategy has not been characterized.

Methods: We used population-level longitudinal data from province-wide registries including plasma viral load, CD4 count, drug resistance, HAART use, HIV diagnoses, AIDS incidence, and HIV-related mortality. We fitted two Poisson regression models over the study period, to relate estimated HIV incidence and the number of individuals on HAART and the percentage of virologically suppressed individuals.

Results: HAART coverage, median pre-HAART CD4 count, and HAART adherence increased over time and were associated with increasing virological suppression and decreasing drug resistance. AIDS incidence decreased from 6.9 to 1.4 per 100,000 population (80% decrease, p = 0.0330) and HIV-related mortality decreased from 6.5 to 1.3 per 100,000 population (80% decrease, p = 0.0115). New HIV diagnoses declined from 702 to 238 cases (66% decrease; p = 0.0004) with a consequent estimated decline in HIV incident cases from 632 to 368 cases per year (42% decrease; p = 0.0003). Finally, our models suggested that for each increase of 100 individuals on HAART, the estimated HIV incidence decreased 1.2% and for every 1% increase in the number of individuals suppressed on HAART, the estimated HIV incidence also decreased by 1%.

Conclusions: Our results show that HAART expansion between 1996 and 2012 in BC was associated with a sustained and profound population-level decrease in morbidity, mortality and HIV transmission. Our findings support the long-term effectiveness and sustainability of HIV treatment as prevention within an adequately resourced environment with no financial barriers to diagnosis, medical care or antiretroviral drugs. The 2013 Consolidated World Health Organization Antiretroviral Therapy Guidelines offer a unique opportunity to further evaluate TasP in other settings, particularly within generalized epidemics, and resource-limited setting, as advocated by UNAIDS.

Citation: Montaner JSG, Lima VD, Harrigan PR, Lourenço L, Yip B, et al. (2014) Expansion of HAART Coverage Is Associated with Sustained Decreases in HIV/AIDS Morbidity, Mortality and HIV Transmission: The ‘‘HIV Treatment as Prevention’’ Experience in a Canadian Setting. PLoS ONE 9(2): e87872. doi:10.1371/ journal.pone.0087872

Editor: Nicolas Sluis-Cremer, University of Pittsburgh, United States of America

Received November 5, 2013; Accepted December 30, 2013; Published February 12, 2014

Copyright: � 2014 Montaner et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: The work here described [Treatment as Prevention in BC initiative (PI: JM)] has been financially supported by the British Columbia Ministry of Health, as well as an Avant-Garde Award (No. 1DP1DA026182) and grant 1R01DA036307-01 from the National Institute of Drug Abuse (NIDA), at the US National Institutes of Health (NIH). In addition, JM was funded through a Knowledge Translation Award from the Canadian Institutes of Health Research (CIHR). VDL is supported by a Scholar Award from the Michael Smith Foundation for Health Research and a New Investigator Award from CIHR, and by a NIDA award (R03DA033851) and CIHR award (MOP-125948). TK was supported in part by a NIDA award (R01DA028532) and a CIHR award (MOP–102742). EW is supported by a Tier 1 Canada Research Chair in Inner City Medicine. Additional limited unrestricted funding was provided by Abbvie, Boehringer-Ingelheim, Bristol-Myers Squibb, Gilead Sciences, Janssen, Merck and ViiV Healthcare. None of the funders played any role in study design, data collection and analysis, decision to publish, preparation or review of the manuscript.

Competing Interests: This study received limited unrestricted funding from Abbvie, Boehringer-Ingelheim, Bristol-Myers Squibb, Gilead Sciences, Janssen, Merck and ViiV Healthcare. The funding provided by the commercial sources has been directed to the institution and not to the investigators. There are no patents, products in development or marketed products to declare. This does not alter the authors’ adherence to all the PLoS ONE policies on sharing data and materials.

* E-mail: jmontaner@cfenet.ubc.ca

PLOS ONE | www.plosone.org 1 February 2014 | Volume 9 | Issue 2 | e87872

Introduction

After 30 years, controlling the HIV epidemic remains an

extraordinary challenge. This is despite the availability of a

number of proven prevention tools, including harm-reduction

strategies and emerging biomedical interventions. [1–7] Recently,

increasing attention has been focused on the potential role that the

expansion of HIV treatment may offer to curb progression to

AIDS and premature death among HIV infected individuals and

secondarily reduce HIV transmission, commonly referred to as

HIV treatment as prevention or TasP. [8–11] In brief, HIV-1

RNA concentration (hereafter referred to as viral load) is a key

determinant of the level of risk associated with sexual, vertical and

needle sharing-related HIV transmission. [12–21] Appropriate use

of HAART suppresses HIV replication on a sustained basis,

leading typically to undetectable, viral load in plasma and halting

disease progression to AIDS and premature death.11 In addition,

as viral load rapidly declines in plasma and subsequently in other

biological fluids (including semen, vaginal fluids and rectal

mucosa), the likelihood of HIV transmission per exposure event

is markedly reduced. [12–21] The concept of scaling up highly

active antiretroviral therapy (HAART), commonly referred to as

TasP, has gained substantial momentum, as its efficacy and

effectiveness have become increasingly apparent. [8,9,22–24]

However, the real-world population-level effectiveness and

sustainability of this strategy remains to be adequately character-

ized.

British Columbia (BC), Canada, provides a unique environment

to address this issue within a concentrated HIV epidemic.

HAART eligibility in BC has remained consistent with the IAS-

USA guidelines since 1996 to the present. [11,25] BC is an

adequately resourced environment with a publically funded health

care system, that fully subsidizes access to medical services,

centralized laboratory monitoring and access to HAART with no

co-payments or deductibles. In addition, a single agency, the BC

Centre for Excellence in HIV/AIDS (BC-CfE), is responsible for

the centralized distribution of all antiretrovirals and monitoring of

key HIV-related outcomes in BC. Additionally, the availability of

unique personal health numbers for all BC-residents provide

unique opportunities to evaluate the impact of this strategy

throughout the province using anonymized data linkages between

administrative datasets.

We therefore conducted a longitudinal ecological study to

evaluate the population-level effectiveness and sustainability of

HAART expansion in BC. Specifically, we sought to characterize

the association between HAART coverage, and the proportion of

individuals virologically suppressed with the number of new AIDS

diagnoses, and all-cause mortality among HIV-infected BC

residents, as well as the number of new HIV diagnoses and

estimated HIV incident cases between 1996 and 2012.

Methods

Aggregate- and individual-level data on the key study variables

were collected from a number of sources, including the BC-CfE

Laboratory and Drug Treatment Registries, the BC Centre for

Disease Control (BCCDC), the BC Ministry of Health (BC MoH)

population-level health resource utilization files, and the Public

Health Agency of Canada (PHAC).

HIV Incidence and Prevalence We also obtained HIV prevalence estimates for BC from 1996

to 2011, independently generated by the Public Health Agency of

Canada (PHAC) [26] using previously published methods. [27]

The method is a modified back-projection method to estimate

HIV incidence and prevalence. Unlike the back-projection

methods used the literature, this new method does not require

linking HIV and AIDS diagnostic registries. It is based on linking

the estimated parametric distribution between the time to HIV

testing and time since HIV infection. These distributions were

adjusted for testing practices over time (e.g. HIV testing trends),

reporting delays, multiple reporting of cases, AIDS cases reporting

(adjusting for the effect of HAART), survival time before and after

HAART was introduced, and birth cohort effects. [28,29] Data for

2012 are not yet available, and therefore, we extrapolated the

available data to obtain preliminary estimates for these two

indicators for the year 2012.

New HIV Diagnoses and AIDS Morbidity and Mortality The BCCDC collates all BC HIV surveillance data and houses

HIV testing data from the provincial public health reference

laboratory, which accounts for approximately 90% of all HIV

screening and all confirmatory testing in BC. Of note, HIV

infection became provincially reportable in 2003.30 Data regard-

ing new HIV diagnoses were extracted from annual and monthly

BCCDC HIV/AIDS Update and Reportable Diseases reports

from 1996 to 2012. [30] The AIDS case-reports were allocated

according to the year when a client was diagnosed with the first

AIDS defining illness.

HIV Treatment and Monitoring HIV/AIDS care is fully publically funded by the health care

system in BC. This includes publically funded access to medical

services, virological and laboratory monitoring and HAART (with

no co-payments or deductibles). HAART use in BC is driven by

the BC-CfE HIV treatment guidelines, [31] which have remained

consistent with those of the International AIDS Society-USA (IAS-

USA) since 1996. [11,25].

Clinical, treatment and laboratory data were obtained from the

BC-CfE, which has a centralized system capturing all antiretro-

viral distribution, all plasma viral load testing, and all resistance

testing, as well as baseline CD4 count for 96% of all patients

starting antiretroviral therapy in BC. Specifically, HAART

coverage, CD4 cell counts, viral load, treatment adherence (as

measured by validated prescription refill compliance), [32] and

genotypic drug resistance were obtained from the BC-CfE

databases. For HAART coverage, we obtained the yearly number

of individuals on HAART in BC. Viral Load data were adjusted

for changes in viral load assay sensitivity over time, as previously

described. [33] For the purpose of these analyses, we recorded the

highest viral load for every individual per year. To accommodate

irregular frequency of measurements or missing values, the highest

yearly value was carried forward until a new measurement was

available. Individuals were censored if they moved out of the

province or died.

Population-level adherence to antiretroviral therapy was assess-

ed by prescription refill compliance for each year as previously

described. [32] In brief, this was estimated by dividing the number

of months of medications dispensed by the number of months of

follow-up during each calendar year. Resistance testing was

performed on stored viral load samples collected immediately

before starting HAART and upon virological rebound thereafter.

Samples were assigned to 1 of 4 resistance categories on the basis

of a modified International AIDS Society-USA table, as previously

described. [34] In brief, samples were considered to be resistant if

they displayed one or more major resistance mutations in any of

the following 4 categories: I) lamivudine, or emtracitabine; II)

other nucleoside reverse transcriptase inhibitors; III) protease

TasP Experience in Canada

PLOS ONE | www.plosone.org 2 February 2014 | Volume 9 | Issue 2 | e87872

inhibitors; and IV) non-nucleoside reverse-transcriptase inhibitors.

Then, the data were classified into the following categories: I)

individuals with no evidence of HIV resistance or with wild type

HIV only II) those who were never genotyped, and individuals

demonstrating HIV resistance to III) one, IV) two or V) three

antiretroviral drug classes, respectively. Those who were never

genotyped include individuals with viral load samples ,250 copies/mL, which were not tested given that sequence analysis is

not reliable in this setting. Antiretroviral resistance data are

displayed by calendar year, and include all individuals who have

ever enrolled on the BC-CfE program – whether or not they ever

used antiretrovirals, from January 1996 to December 2012.

All-Cause Mortality among HIV-Positive Individuals HIV-related mortality data were obtained from BC Vital

Statistics for all HIV-positive individuals regardless of whether or

not they had or not ever accessed antiretroviral therapy between

1996 and 2011. [35] Data for 2012 were not yet available. The

HIV-related diagnoses were based on the 10th revision of the

International Statistical Classification of Diseases and Related

Health Problems (ICD-10) codes B20 to B24. [36].

Population Data Population estimates were obtained for the years 1996–2011 to

calculate mortality and AIDS rates. These estimates were obtained

through Statistics Canada through BCStats. [37] Data for 2012

were not yet available.

Statistical Analysis To be consistent throughout the text, we included the year 2012

for all variables. However, since the estimates of the Public Health

Agency of Canada were complete until 2011, we used a

polynomial function of degree three to extrapolate the data. This

polynomial function provided the best fit for the incidence and

prevalence data from 1996 to 2011, and we therefore used

estimated values of these two indicators for 2012.

We modeled trends using generalized additive models, which

accounts for the non-linear temporal trends in these longitudinal

data. [38,39] We fitted different Poisson regressions, [40] with

correction for over dispersion in the data, for each of the following

outcomes: estimated HIV incidence rate, HIV-related mortality

rate and the rate of AIDS cases. The primary explanatory

variables of interest were the number of individuals actively on

HAART and the percentage of individuals with the highest yearly

viral load lower than 500 copies/mL. Note that for the rates, the

size of the BC population was used as an offset in the models.

These models were performed using the statistical package SAS

(version 9.3). All p values reported were two-sided, and

significance was set at the 5% level.

Ethics Approval The BC-CFE received approval for this study from the

University of British Columbia ethics review committee at the St

Paul’s Hospital, Providence Health Care site (P05–123). The study

complies with the BC’s Freedom of Information and Protection of

Privacy Act. The study was conducted primarily using anon-

ymized laboratory and administrative databases, and therefore

informed consent was not required. Incidence data were

augmented with data collected through prospective research

cohort studies, which include written informed consent by study

participants and separate IRB approval.

Role of the Funding Source The sponsors had no role in the design, data collection, data

analysis, data interpretation, or writing of the report. The

corresponding author had full access to all data in the study and

had final responsibility to submit for publication.

Results

From January 1st 1996 to December 31st 2012, the estimated

HIV prevalence in BC increased from 7,900 to 11,972 cases (52%;

p-value ,0.0001), and the number of individuals actively on HAART increased from 837 to 6772 (709%; p-value ,0.0001). Based on these figures, we estimated that HAART coverage

increased from 11% to 57% (p-value 0.0004) during this period.

HIV Disease Progression As shown in Figure 1, the trend in the number of HIV-related

deaths from 1996 to 2011, with 253 individuals dying in 1996 and

only 59 individuals dying in 2011. Using the overall BC population

as a denominator, the HIV-related mortality rate decreased from

6.5 to 1.3 per 100,000 of the BC population during 1996–2011

(80% decrease, p-value 0.0115). Figure 1 also shows that the AIDS

rates decreased from 6.9 to 1.4 per 100,000 population over the

study period (80% decrease, p-value 0.0330).

As shown in Figure 2, the overall median baseline or pre-

therapy CD4 cell count increased from 270 cells/mL (25–75th

percentile 130–390) in 1996 to 380 cells/mL (25–75th percentile 235–550) in 2012 (41% increase; p-value ,0.0001). The median pre-therapy CD4 cell count among individuals with a history of

injection drug use (IDU) was 350 cells/mL (25–75th percentile 180–500) in 2012.

As shown in Figure 3, HAART adherence levels for individuals

who ever started therapy in BC increased over time. While only

37% of individuals had adherence levels $95% in 1996, this increased to 71% for individuals (p-value 0.0032) by 2012.

Table 1 shows the estimated aggregate viral load levels at the

population-level over time. Using a 500 copies/mL cut-off, we

estimated that the proportion of individuals virologically sup-

pressed increased from 8% in 1996 to 74% in 2012 (p-value , 0.0001). The analysis was repeated using a 50 copies/mL cut-off;

however, this was restricted to the period of 1999–2012, as the

more sensitive test was not available prior to 1999. Using a 50

copies/mL cut-off, the proportion of individuals suppressed

increased from 6% to 59% (p-value ,0.0001). As shown in Figure 4, emergent antiretroviral drug resistance

among individuals with unsuppressed viral load (including

individuals on or off therapy) decreased from 1996 to 2012 in

BC. Since 1996, the number of individuals never genotyped

decreased markedly from 64% in 1996 to 10% in 2012 (p-value , 0.0001). At the same time, the prevalence of individuals with wild

type virus (i.e.: no drug resistance) increased from 15% in 1996 to

69% in 2012 (p-value ,0.0001).

HIV New Diagnoses As shown in Figure 5, there was a steady decline in the number

of HIV new diagnoses from 702 to 238 cases (266%; p-value 0.0004) and for estimated reduction of HIV incident cases from

632 to 368 cases per year (242%; p-value 0.0003) between 1996 and 2012. New HIV diagnoses decreased by 92% (p-value 0.0013)

among individuals with a history of injection drug use, and by 22%

(p-value 0.0046) among MSM. Further, when the size of the MSM

population was factored in, based on US CDC estimates, the rates

of HIV new diagnoses was found to have declined from 4.43 per

1000 in 1996 to 3.21 per 1000 in 2004 to 1.81 per 1000 in 2012.

TasP Experience in Canada

PLOS ONE | www.plosone.org 3 February 2014 | Volume 9 | Issue 2 | e87872

All confirmed HIV positive tests have been documented since

1989, however HIV became reportable in 2003 in BC with

systematic follow-up of positive test results from that time forward.

Therefore reports prior to 2003 are biased because duplicate

positive tests were not consistently removed. All longer-term trends

in Figure 5 after 2003 are statistically significant. Therefore it is

unlikely that over-counting cases prior to 2003 substantially

impacted the overall conclusions of the paper.

Statistical Models Next, we developed two statistical models with the outcome

being the estimated HIV incidence rate and the explanatory

variables being the number of individuals actively on HAART and

the percentage of individuals suppressed on HAART (using a 500

copies/mL cut-off to allow a consistent definition of suppression

over time). The model showed that for each 100 individuals

actively on HAART the estimated incidence rate decreased by

1.2% (estimated rate ratio 0.9879; 95% CI 0.9868; 0.9891) and for

each 1% increase in the number of individuals suppressed on

HAART, the HIV incidence decreased by 1% (estimated rate

ratio 0.9900; 95% CI 0.9887; 0.9912).

For the model with HIV-related mortality rate as the outcome

and the explanatory variables being the number of individuals

actively on HAART and the percentage of individuals suppressed

on HAART (using a 500 copies/mL cut-off), we observed that for

each 100 individuals actively on HAART the estimated mortality

rate decreased by 2.51% (estimated rate ratio 0.9749; 95% CI

0.9703; 0.9795) and for each 1% increase in the number of

individuals suppressed on HAART, the mortality rate decreased

by 2.06% (estimated rate ratio 0.9794; 95% CI 0.9754; 0.9834).

For the model with the outcome being the AIDS rate and the

explanatory variables being the number of individuals actively on

HAART and the percentage of individuals suppressed on HAART

(using a 500 copies/mL cut-off), we observed that for each 100

individuals actively on HAART the estimated AIDS rate

decreased by 2.48% (estimated rate ratio 0.9752; 95% CI

0.9679; 0.9826) and for each 1% increase in the number of

individuals suppressed on HAART, the AIDS rate decreased by

1.95% (estimated rate ratio 0.9805; 95% CI 0.9737; 0.9874).

Discussion

Our results demonstrate that between 1996 and 2012, the

expansion of HAART coverage in BC was strongly and

statistically significantly associated with population level decreases

in incident AIDS diagnoses and all cause mortality among HIV

infected individuals. In addition, the expansion of HAART

coverage in BC was strongly and statistically significantly

associated with decreased HIV new diagnoses and estimated