Monitoring of HIV treatment in seven countries in the WHO Region of the Americas

Abstract Objective To determine the prevalence of adequate monitoring and the costs of measuring CD4+ T-lymphocytes (CD4+ cell) and human immunodeficiency virus (HIV) viral load in people receiving antiretroviral therapy (ART) in seven countries in the WHO Region of the Americas. Methods We obtained retrospective, longitudinal data for 14 476 adults who started a first ART regimen at seven HIV clinics in Argentina, Brazil, Chile, Haiti, Honduras, Mexico and Peru between 2000 and 2011. We estimated the proportion of 180-day periods with adequate monitoring, which we defined as at least one CD4+ cell count and one viral load measurement. Factors associated with adequate monitoring were analysed using regression methods. The costs of the tests were estimated. Findings The median follow-up time was 50.4 months; the proportion of 180-day periods with adequate CD4+ cell counts was 69% while the proportion with adequate monitoring was 62%. Adequate monitoring was more likely in participants who were older, who started ART more recently, whose first regimen included a non-nucleoside reverse transcriptase inhibitor or who had a CD4+ cell count less than 200 cells/µl at ART initiation. The cost of one CD4+ cell count ranged from 7.37 United States dollars (US$) in Argentina to US$ 64.09 in Chile; the cost of one viral load measurement ranged from US$ 20.34 in Brazil to US$ 186.28 in Haiti. Conclusion In HIV-infected participants receiving ART in the WHO Region of the Americas, CD4+ cell count and viral load monitoring was often carried out less frequently than regional guidelines recommend. The laboratory costs of monitoring varied greatly.


Introduction
For people living with human immunodeficiency virus (HIV) who are receiving combination antiretroviral therapy (ART), the most important predictors of treatment outcomes are CD4+ T-lymphocyte (CD4+ cell) count and HIV load in the blood. 1 Both international and national health organizations recommend that the CD4+ cell count, viral load or both be measured routinely every three to 12 months during ART. [2][3][4] In Latin America, national clinical guidelines recommend measuring the CD4+ cell count and the viral load every two to six months in people starting ART and then every three to six months once viral suppression has been achieved. [5][6][7][8][9][10] Recently, the World Health Organization (WHO) updated its guidelines to promote earlier treatment initiation and enhanced monitoring, preferably by viral load testing. 2 The number of people receiving ART and the cost of treatment are expected to increase in the future because of the worldwide trend to initiate ART at higher CD4+ cell counts and earlier identification of HIV-infected individuals. Improvements in life expectancy resulting from expanded ART programmes are also thought to increase costs. 11 WHO has been working with regional United Nations agencies and in-country staff and with national ministries of health to sup-port the development and implementation of ART guidelines in individual countries. 12 At present, however, there is little information on the application of, and adherence to, current WHO and national clinical guidelines in the WHO Region of the Americas. The aims of this study were to determine how frequently the CD4+ cell count and HIV viral load were monitored in people receiving ART in the region between 2000 and 2011. We also wanted to assess the level of adherence to local clinical guidelines and to identify factors associated with infrequent CD4+ cell count and viral load monitoring. We also estimated the cost of CD4+ cell count and viral load measurements in the region.

HIV treatment monitoring in the Americas
Pablo F Belaunzarán-Zamudio et al.
and 31 December 2011 were eligible for inclusion. We excluded participants enrolled in clinical trials to avoid potential bias due to special monitoring practices in these trials. The follow-up period after ART initiation was divided into 180-day periods; follow-up ended on the day of the last recorded visit or when death occurred. Participants were defined as being lost to follow-up if they had not visited the clinic in the 12 months before the database closing date of 1 January 2012. For the primary analysis, treatment monitoring was defined as adequate if at least one CD4+ cell count and one viral load measurement had been made in each 180-day follow-up period. The site in Haiti did not measure viral load during the study period and was not included in the primary analysis. However, the Haitian site was included in a secondary analysis, in which we defined adequate monitoring using the CD4+ cell count alone (in which case, at least one CD4+ cell count had to be measured during each 180-day period).
The primary study outcome was the proportion of 180-day periods during which there was adequate monitoring. Any remaining follow-up time that did not fit within a 180-day period was ignored. For example, if a participant was followed for 400 days, two 180-day periods were included in calculating the proportion of follow-up with adequate monitoring; the remaining 40 days were not included. However, the remaining follow-up periods were included in sensitivity analyses as additional periods. The CD4+ cell count at ART initiation was defined as the measurement closest to the start of ART, but no more than 180 days before or 7 days after the start, and was categorized as either less than 200, 200-350 or more than 350 cells/µL. A participant was described as having had an AIDS-defining event before ART initiation if they had clinical disease that could be classified as Centers for Disease Control and Prevention category C or WHO stage IV. Combination antiretroviral therapy was defined as: (i) therapy based on a non-nucleoside reverse transcriptase inhibitor (e.g. one non-nucleoside reverse transcriptase inhibitor plus two nucleoside reverse transcriptase inhibitors); (ii) protease inhibitor-based therapy, including treatment with one ritonavir-boosted or unboosted protease inhibitor plus two nucleoside reverse transcriptase inhibitors; (iii) triple nucleoside reverse transcriptase inhibitor regimens; or (iv) any  HIV treatment monitoring in the Americas Pablo F Belaunzarán-Zamudio et al.
other regimen containing at least three drugs. Data were collected at each study site and entered into the database at the local level using codes that did not identify individual participants. Thereafter, data were sent for harmonization to the CCASAnet data coordinating centre at Vanderbilt University, Nashville, United States of America. The coordinating centre also carried out data quality checks and on-site audits to ensure data accuracy. Ethical approval was obtained from institutional review boards at each study site and at Vanderbilt University.

Statistical analysis
Participant-level factors associated with the frequency of monitoring were identified using Poisson regression: the number of 180-day periods with adequate monitoring was the outcome and the total number of 180-day periods was the offset. A quasi-Poisson estimation procedure was used. Multivariable models included age, sex, CD4+ cell count at ART initiation, year of ART initiation, prior AIDS-defining events and the first ART regimen. Missing data on the CD4+ cell count and prior AIDS-defining events were imputed using multiple imputation with 10 replications. The rate ratio for adequate monitoring was computed for each site and the combined rate ratio across all sites was estimated using random effects meta-analysis. The costs of measuring the CD4+ cell count and viral load were obtained for each site in 2014 in the local currency. Costs were subsequently converted to United States dollars (US$) using exchange rates from the Wall Street Journal for Argentina, Brazil, Chile, Mexico and Peru 15 and rates from local central banks for Haiti 16 and Honduras. 17 The median cost of all CD4+ cell counts and viral load measurements per participant per year were estimated for each site. All analyses were performed using R statistical software (R-foundation, Vienna, Austria). Analysis scripts are available from the corresponding author.

Frequency and cost
The frequency and cost of CD4+ cell counts and viral load measurements varied according to the study site ( Table

Adherence to guidelines
The adequacy of CD4+ cell count and viral load monitoring is shown for each site in Fig. 1. The proportion of periods with adequate monitoring was highest in Mexico (86%) and Argentina (80%) but much lower in Honduras (26%). Across the six sites for which data were available, the proportion of periods with adequate CD4+ cell count and viral load monitoring was 62% (95% confidence interval, CI: 52-73) -the large confidence interval was due to the substantial variation between sites. Fig. 2 shows the proportion of 180-day periods with adequate CD4+ cell count monitoring alone. Overall, the proportions were higher than those observed for CD4+ cell count and viral load monitoring combined and ranged from 86% and 81% in Mexico and Argentina, respectively, to 54% and 48% in Honduras and Haiti, respectively. Across all seven sites, the proportion of periods with adequate CD4+ cell count monitoring alone was 69% (95% CI: 57-82). General trends were similar in sensitivity analyses that included periods of less than 180 days (data available from corresponding author).
Factors associated with adequate CD4+ cell count and viral load monitoring at each site and across all sites combined were identified. At all sites, adequate CD4+ cell count and viral load monitoring was more likely in older participants (Fig. 3) and in those who started ART more re-cently (Fig. 4). Patients with a CD4+ cell count less than 200 cells/µL at ART initiation were more likely to have adequate monitoring than those with a count more than 350 cells/µL (Fig. 5). Participants whose first ART regimen contained a nonnucleoside reverse transcriptase inhibitor were also more likely to have adequate monitoring than those treated with other regimens (Fig. 6). Neither sex, a CD4+ cell count in the range 200 to 350 cells/µL nor prior AIDS-defining events influenced the likelihood of adequate CD4+ cell count and viral load monitoring.  HIV treatment monitoring in the Americas Pablo F Belaunzarán-Zamudio et al.
Factors associated with adequate CD4+ cell count monitoring alone were also identified. Again, adequate monitoring was more likely in older participants (Fig. 7) and in those whose first ART regimen contained a nonnucleoside reverse transcriptase inhibitor (Fig. 8). Sensitivity analyses that included periods shorter than 180 days yielded similar results (details available from the corresponding author). Neither sex, prior AIDS-defining events, nor the year of ART initiation influenced the likelihood of adequate CD4+ cell count monitoring, though there was some indication that a CD4+ cell count less than 200 cells/µL at ART initiation may have had an effect.

Discussion
We found that HIV-infected people starting ART at seven sites in the region were monitored less frequently than recommended by regional and national clinical guidelines. On average, CD4+ cell count and viral load monitoring were done at least once every six months for only 62% of the time participants were in care, although the proportion varied greatly across sites. The equivalent proportion for CD4+ cell count monitoring alone was 69%. Adequate monitoring of both parameters was associated with older age at ART initiation, receiving ART in more recent years, starting ART when the CD4+ cell count was less than 200 cells/µL and a first ART regimen that included a non-nucleoside reverse transcriptase inhibitor.
The annual cost of CD4+ cell count and viral load monitoring per participant varied greatly between countries. For instance, the annual cost of CD4+ cell count monitoring per participant in Mexico was around four times the median cost in the region and almost nine times the cost in Honduras. Although we are not aware of previous studies of the cost of monitoring ART efficacy in the region, the Pan American Health Organization (PAHO) recently noted that the annual cost of ART medications per participant in the most expensive countries was approximately 10 times that in the lowest-cost countries. 18 These differences have been attributed to the varying use of generic drugs and to the ART procurement mechanism preferred by each country. 18 , 19 The implications of the variability in monitoring costs across the region we observed are difficult to discern. For instance, although it appears that the cost of CD4+ cell count and viral load monitoring at individual sites was not related to the frequency of monitoring or to adherence to guideline recommendations, we were not able to assess how cost differences affected participant care, such as the ability to detect treatment failure. Furthermore, it would be diffi-cult to use our results to identify a single initiative that could reduce costs across the region because each centre itself negotiates costs and budgets with local laboratories and with different sponsors. For the large centres in Argentina, Brazil, Chile and Peru, it may be possible to reduce costs by exploiting the large demand to leverage services. However, this strategy may not work for centres in Haiti, Honduras or Mexico, where government and international funding The frequency of CD4+ cell count and viral load monitoring we observed in HIV-infected persons receiving ART was very similar to that reported in the region by PAHO. 20 It is possible to divide our sites into two groups: those where participants were followed up in accordance with clinical guidelines most of the time (i.e. Argentina, Brazil, Chile, Mexico and Peru) and those where participants were followed up in accordance with guidelines less than half the time (i.e. Haiti and Honduras). It is noteworthy that the proportion of periods when CD4+ cell count and viral load monitoring was adequate was very similar to the proportion when CD4+ cell count monitoring alone was adequate at all sites except Haiti, which did not measure viral load, and Honduras, where in practice laboratory monitoring of the efficacy of ART was based more on CD4+ cell counts alone, than on both CD4+ cell count and viral load monitoring.

Fig. 5. Adequate CD4+ cell count and HIV viral load monitoring in six countries in the WHO Region of the Americas, 2000-2011: rate ratio for CD4+ T-cell count at ART initiation >350 cells/ µL versus <200 cells/µL
The low level of adherence to clinical guidelines we observed was probably due to a combination of factors related to the participants, physicians and other health-care providers and, more generally, to structural and programmatic characteristics. At the individual level, our findings agree with those of previous studies in identifying older age as an important factor associated with adherence to ART, 21 adequate CD4+ cell count monitoring, retention on treatment and fewer missed clinic visits. 22 , 23 We found that a CD4+ cell count less than 200 cells/µL at ART initiation was associated with adequate CD4+ cell count monitoring overall, but not at most individual sites. We hypothesize that clinicians' perception that profound immunosuppression is associated with an increased risk of complications may have led to more frequent monitoring.
Although we were not able to evaluate structural and programmatic characteristics in detail, we observed several factors that could explain some of the differences in monitoring between sites. For example, the Global Fund to Fight AIDS, Tuberculosis and Malaria did not cover the cost of viral load measurements in Haiti and, consequently, this test was little used.
Haiti and Honduras have generalized HIV epidemics, whereas in the rest of the region HIV infection is concentrated among men who have sex with men and injection-drug users. 24 There were also clear differences in national income, development indices and health expenditure per capita 25 between Haiti and Honduras and the other countries we studied. 25      Our study has several limitations. First, we used a conservative definition of adequate CD4+ cell count and viral load monitoring -six months is currently the maximum time interval recommended by the guidelines used at all sites and most countries in the region recommend measurements every three to four months. 5 -10 Thus, we may have overestimated adherence to recommendations. Since we were not able to identify reasons for poor adherence or for the wide range in monitoring costs, it is difficult to propose interventions that would improve adherence or decrease costs. Another limitation is that it is not clear whether poor adherence to laboratory monitoring guidelines was associated with worse outcomes in HIV-infected people in the region. Certainly, virological failure will be detected less often if fewer viral load measurements are done and this may lead to poorer clinical outcomes. Earlier research at our study sites found that mortality was generally higher in Haiti and Honduras, 28 where monitoring was less frequent, but the high mortality was probably due to several factors and not to monitoring practices alone. Previous studies in Africa and mathematical modelling suggest that carrying out laboratory monitoring less frequently than recommended may have negative consequences for the health of people on ART and that measuring the viral load two or three times a year may be more cost-effective and reduce HIV transmission. 29 , 30 However, the clinical and cost benefits of measuring both the CD4+ cell count and viral load as frequently as currently recommended are still debated. 31 -35 A randomized trial may be required to resolve the issue.
In conclusion, adherence to recommended CD4+ cell count and viral load monitoring frequencies was generally poor in the region, with large variations between sites. Laboratory costs were also highly variable. Further research is needed to identify the underlying cause of differences in the frequency and cost of monitoring across the region and studies should be carried out to evaluate the impact of less frequent laboratory monitoring on health outcomes in HIVinfected persons receiving ART. We hope our study findings will be helpful for developing and implementing HIV treatment guidelines. ■

HIV treatment monitoring in the Americas
Pablo F Belaunzarán-Zamudio et al.