Managing HIV-infected patients on antiretroviral therapy in Rio de Janeiro, Brazil: do providers follow national guidelines?

The objective of the paper was to compare provider practices in prescribing antiretroviral (ARV) drug regimens and use of laboratory monitoring at three health care facilities and to determine whether Brazilian national guidelines are being followed. A retrospective, cross-sectional survey was employed. We selected a sequential sample of patients on ARV therapy who registered at three health care facilities in Rio de Janeiro, Brazil, during 2001. We abstracted 2001 patient visit data from medical records using standardized data forms. Provider practice was compared to the 2000 Brazil national guidelines for ARV use. Providers who prescribed recommended or acceptable regimens were considered as having conformed to guidelines. Only 2% of patient records (N=984) reported use of inappropriate regimens as defined by the Brazil 2000 national ARV guidelines. Forty-nine per cent of patients at the Evandro Chagas hospital, 17% of those at Hospital Geral, and 57% of those at Centro da Saude were prescribed recommended therapies. Twenty per cent of patients seen at the public district hospital received dual ARV therapy, an acceptable regimen at the time. Although the national guidelines do not provide recommendations on laboratory monitoring, during the 1 year study period a majority of patients had at least one CD4+ cell count (92%) or viral load measurement (86%). Providers' practices in prescribing ARV regimens at these Rio de Janeiro facilities conform to national guidelines. Physicians would benefit from Brazilian ARV guidelines which incorporate the international consensus on the frequency of laboratory monitoring appropriate for patients in resource-constrained settings.

Since 1996, the Brazilian government has provided free access to antiretroviral (ARV) medications, including protease inhibitors, for all clinically eligible persons in the country with HIV infection. By 2001, 105,000 patients had received ARV medication through the public health system (Sudo et al.,[5]). In addition to access to ARVs, the government programme provides CD4+ cell count and viral load testing at no cost to patients.

Since 1996, the National AIDS and STD Program of the Ministry of Health has coordinated annual meetings of the Advisory Group on Antiretroviral Therapy to revise the national ARV guidelines (Ministerio da Saude, [4]). The recommendations largely conform to those of the International AIDS Society—USA Panel (Carpenter et al., [2]) with the exception that initiation of two nucleoside reverse transcriptase inhibitors (NRTIs) was still considered acceptable in the Brazilian guidelines up to 2000.

Using patient medical records from 2001, we evaluated whether national ARV guidelines were being followed in three different types of health facilities in Rio de Janeiro, Brazil.

We conducted a cross-sectional study based on a retrospective review of medical records of adult patients (>13 years of age) from three government health facilities in Rio de Janeiro, Brazil: the Instito de Pesquisa Clinica Evandro Chagas (referred to as Evandro Chagas), a specialized infectious diseases hospital; the Hospital Geral Nova Iguaçu (referred to as Hospital Geral), a federal public reference hospital located in a poor suburb; and the Centro da Saude Dr Vasco Barcelos (referred to as Centro da Saude), a public ambulatory care facility also located in Nova Iguaçu. Evandro Chagas and Hospital Geral offered their ARV patients on-site CD4+ cell count and viral load testing. Patients at Centro da Saude could obtain CD4+ cell counts on-site but were referred to Hospital Geral for viral load testing.

The sample size calculated was adequate to detect with 80% power and α=0.05 at least a 15% difference in the proportion of patients receiving recommended regimens between facilities.

We selected patients on ARV therapy who received care at the facility in 2001, starting with those appearing most recently in the patient logs or pharmacy records. Some of these patients began ARV therapy prior to 2001.

Trained staff abstracted data from the medical records from 2001 using a form developed for this evaluation (Carmody et al., [1]). The variables collected were date of birth, gender, level of education, exposure category associated with HIV infection, the clinical status of the patients, the date of AIDS diagnosis, hospitalization, or death, the date and result of any CD4 + cell count or viral load measurement performed, the date of prescription, and the combination of drugs prescribed.

We used three categories to define whether a prescribed regimen followed the 2000 Brazilian national guidelines for ARV use: 'recommended' for those combinations listed explicitly as initial regimens; 'inappropriate' for those regimens listed as contraindicated; and 'acceptable' for those regimens which were not listed in the guidelines as either recommended or inappropriate. Of note, the 2000 guidelines listed certain dual therapy regimens as recommended. We considered physicians as conforming to guidelines if they prescribed recommended or acceptable regimens.

Both the Brazil national guidelines and international consensus guidelines (Ministerio da Saude, [4]) recommend the general use but do not specify the frequency of laboratory monitoring of patients' immunological and virological response to treatment.

We used χ2 tests to identify statistically significant overall differences in distributions of categorical variables. Fisher's exact tests were used when small cell size made χ2 tests inappropriate. All analyses were conducted using Statistical Analysis Systems (SAS) version 8 (SAS Institute).

This work was determined to be non-research by the Centers for Disease Control Internal Review Board because the abstraction of data from existing patient records was evaluative in nature and intended to directly inform programme improvement.

We abstracted data from the medical records of 984 patients receiving ARV therapy in 2001: 509 from Evandro Chagas, 345 from Hospital Geral, and 130 from Centro da Saude. At each site, a majority of patients were male (Table 1). Most patients had completed primary school. Sexual risk behaviours comprised the primary modes of HIV transmission: 55% to heterosexual exposure and 29% to homosexual exposure.

Table 1. Demographic and clinical characteristics of HIV-infected patient sample from three health facilities in Rio de Janeiro, Brazil, 2001

Less than 20% of patients at Evandro Chagas and Hospital Geral initiated ARVs during the study period, compared to 92% of patients enrolled from Centro da Saude. The three health care facilities also differed in the timing of patients' AIDS diagnosis. The vast majority (96%) of patients at Hospital Geral were diagnosed before 2001, compared with 20% of patients at Evandro Chagas and 3% of patients at Centro da Saude.

A small proportion of patients (0.3%) included in the sample were known to have died during the observation period. A small fraction of patients at Evandro Chagas and Hospital Geral required hospitalization at some time during 2001 (7% and 14%, respectively). In contrast, close to one-fifth of patients from the health centre were admitted during 2001.

Physicians prescribed recommended regimens to 49% of patients at Evandro Chagas, to 17% of those at Hospital Geral, and to 57% of those at Centro da Saude (Table 2). Based on the Brazil 2000 guidelines for ARV therapy, only 2% of patients were prescribed inappropriate regimens.

Table 2. Initial ARV regimens prescribed in 2001 to HIV-infected patients at three health facilities in Rio de Janeiro, Brazil (%) (numbers in parenthesis)

Of those who were not prescribed inappropriate regimens, patients at Centro da Saude were significantly more likely to receive a recommended regimen compared to patients at Hospital Geral or Evandro Chagas (p<0.001).

Dual NRTI therapy was a regimen prescribed to 18% of patients at Evandro Chagas, 23% at Hospital Geral, and 17% at Centro da Saude (Table 2). A majority of patients who were prescribed dual therapy at Evandro Chagas (97%) and Hospital Geral (76%) were patients who initiated therapy before 2001, compared with 9% of patients on dual therapy attending Centro da Saude.

In this sample, the top three regimens prescribed by each clinic comprised less than half of all patients receiving ARV therapy.

A majority of patients on ARV therapy at Evandro Chagas (88%) and Centro da Saude (89%) (Table 3) had at least one CD4+ cell count during 2001. Two-thirds of the patients at Centro da Saude who lacked a CD4+ cell count in 2001 were those who initiated care in the last 90 days of 2001 compared to 5% of those at Evandro Chagas.

Table 3. Propotion of persons without documented laboratory tests after initiating therapy in three facilities in Rio de Janeiro, 2001 (%) (numbers in parenthesis)

Viral load measurement was less common than CD4+ cell count measurement across all sites: 85% at Evandro Chagas, 93% at Hospital Geral, and 69% of patients at Centro da Saude. At Centro da Saude, 65% of patients without viral load measures registered in the last 90 days of 2001. Overall, most patients (83%) without a CD4+ cell count also did not have a viral load measured in 2001.

These data quantify the relatively small degree (2% of patients) to which Brazilian physicians at three health care facilities prescribed regimens deemed inappropriate by national guidelines for ARV patient management.

While physician practice appeared consistent with national guidelines, some issues regarding management of patient on ARV therapy emerge from this analysis.

First, 14% of all patients in this sample were prescribed dual therapy. Since 2001, the Brazil national guidelines no longer recommend dual ARV therapy for patients. Similarly, national data show a decrease in prescription of dual therapy from over 60% in January 1997 to 20% in December 2000 (Levi & Vitoria, [3]).

Secondly, we found a small but substantial proportions of patients who did not have any CD4 (8%) or viral load (14%) performed during the study period. By quantifying the proportion of patients managed without laboratory testing, we hope to encourage efforts to address barriers to accessing laboratory tests. The national programme would benefit from future evaluations to identify factors associated with patients not having CD4+ cell count or viral load measured, particularly the impact of no on-site testing. There are some limitations to this evaluation. We could not assess which patients were co-infected with tuberculosis or determine whether these patients received regimens which were consistent with the national guidelines for managing patients with both tuberculosis and HIV infection.

Thirdly, we evaluated compliance of providers in 2001 against the Brazil national guidelines from 2000. Because Brazil revises its ARV guidelines every year, efforts to improve patient care require periodic evaluation of physician practice as guidelines change. Additional studies are needed to examine whether physicians were aware of national guidelines or referred to these guidelines when they had questions about patient care.

Conducting these types of evaluations may be used to identify gaps in national guidelines as well as areas in which physicians may benefit from additional training or more direction in the choice of regimens. Together with ongoing evaluation of treatment failure rates, development of resistant strains, and access to services, assessments of provider practice provide key information for continual improvement of the Brazil national programme.