The benefits of triple drug highly active antiretroviral therapy (HAART) in the management of HIV disease are well established. Through the suppression of plasma HIV-1 RNA, HAART has been shown to improve CD4 cell counts and in turn decrease morbidity and mortality among HIV–infected patients [1,2]. However, a critical question surrounds the optimal time to initiate HAART.
We have recently evaluated rates of disease progression among patients initiating HAART, and found that only patients initiating therapy when their CD4 cell count had declined below 200 × 106 cells/l were at increased risk of disease progression . Although this finding has recently been independently confirmed [4–7], therapeutic guidelines have been conservative in their recommendations on when to start therapy [3,8,9]. These thresholds have been debated, primarily because of concerns regarding limited follow-up and potential for irreversible immune damage at CD4 cell counts below 350 × 106 cells/l. The British HIV Association Guidelines advise clinicians to consider intitiating therapy when the CD4 cell count has declined to within the range of 200–350 × 106 cells/l while taking into account the rate of CD4 cell count decline and HIV-1 RNA level, and immediate initiation of therapy if CD4 cell count has declined to 200 × 106 cells/l . Conversely, the US Public Health Service guidelines currently recommend treatment if the CD4 cell count declines below 350 × 106 cells/l or if plasma viral load is above 55 000 copies/ml . More recently, the International AIDS Society-USA guidelines specifically support the use of a higher than 200 × 106 cells/l threshold with the decision to treat being based on the rate of CD4 decline and the HIV-1 RNA level .
However, prior to the advent of HAART, it was observed that non-immune-based factors, such as physician experience, were strongly associated with survival . Although a role of physician experience in the HAART era has been speculated upon [13–15], the impact of physician experience in the HAART era has not been demonstrated. In addition, although a wealth of studies have demonstrated an association between adherence to antiretroviral therapy and improved viral suppression [16–18], the relationship between adherence to therapy and AIDS-related mortality in the HAART era has not been well-demonstrated.
For the present analysis, we hypothesized that if 200–350 × 106 cells/l represented an irreversible biological threshold that would limit the response to antiretroviral therapy, adjustment for conservative estimates of non-immune-based factors would not have a substantial impact on HIV-related mortality among patients initiating therapy with < 200 × 106 cells/l. The present analysis was therefore conducted to evaluate whether survival rates could be affected by conservative estimates of physician experience and patient adherence.
The HIV/AIDS treatment program
The dispensation of antiretroviral medications in the province of British Columbia has been described in detail elsewhere [3,19]. In brief, the Centre remains the only free source of antiretroviral medications in the province. The Centre's HIV/AIDS Drug Treatment program has received ethical approval from the University of British Columbia Ethics Review Committee at its St Paul's Hospital site, and the program conforms with the province's Freedom of Information and Protection of Privacy Act.
In June 1996 the Centre adopted plasma viral load-driven antiretroviral therapy guidelines, consistent with those put forward by the International AIDS Society – USA . In brief, antiretroviral therapy-naive individuals with plasma viral load > 100 000 copies/ml were offered triple drug regimens (i.e. two nucleosides plus a protease inhibitor or a non-nucleoside reverse transcriptase inhibitor), whereas those with plasma viral loads from 5000 to 100 000 copies/ml were offered dual nucleoside therapy. The Centre guidelines were revised in July 1997 to recommend triple combination therapy for all antiretroviral-naive individuals with plasma HIV-1 RNA levels greater than 5000 copies/ml or CD4 cell counts below 500 × 106 cells/l . The Centre's guidelines recommend further that plasma HIV-1 RNA levels be monitored at baseline, at 4 weeks after starting antiretroviral therapy and every 3 months thereafter. Plasma viral loads were measured using the Amplicor HIV-1 Monitor® Roche Diagnostics, Branchburg, New Jersey, USA).
At the time of enrollment, physicians enrolling an HIV-infected individual must complete a drug request enrollment form, which acts as a legal prescription and supplies baseline information, including past HIV-specific drug history, CD4 cell count, plasma HIV-1 RNA, current drug requests, and enrolling physician data. The primary care physicians are asked to complete a clinical staging form using the World Health Organization (WHO) clinical staging system . For all program participants, a complete prospective profile of antiretroviral therapy is maintained, including the medications dispensed, the dose, the dispensation dates, and the quantity dispensed.
The primary endpoint in this analysis was death. Deaths occurring during the follow-up period were identified on a continuous basis from physician reports and through record linkages carried out with the British Columbia Division of Vital Statistics. Deaths from accidental causes were censored at time of death, and classified as non-events in the primary analysis. However, we also conducted sub-analyses of all cause mortality.
Physician experience and adherence
Physician experience was defined a priori, based on data derived in the pre-HAART era, which found that having a physician who had provided primary care for greater than five HIV-infected patients was associated with improved survival . Based on these earlier findings and our objective to only adjust for conservative estimates of physician experience, we defined a patient as having an HIV-experienced physician if the patient's enrolling primary care physician had previously enrolled six or more patients into the program. As patients were continuously enrolled in the treatment program during the study period, a ‘non-experienced’ physician could become ‘experienced’ by our definition over time, however, the experience level assigned to each patient was fixed and based on the HIV-related experience of their physician at the time of each patient's first follow-up.
Patients were conservatively defined as adherent or non-adherent using prescription refill compliance . The definition of adherence was based on the ratio of time that medication dispensed would last as a proportion of follow-up time. This calculation was restricted to each patient's first year on therapy to avoid reverse causation that could occur among patients who cease antiretroviral therapy after they have become too sick to take medication. If a person died prior to 1 year after starting therapy, then the proportion of follow-up time and the proportion of time receiving treatment were used to calculate this ratio. We have previously demonstrated how this estimate strongly predicts virological response, and how it can adjust for the potentially confounding effect of treatment interruption [24–26]. For the purposes of this analysis, patients were defined as non-adherent if they received antiretroviral medications < 75% of the time during the first year of follow-up, since this level of adherence is well below what is presently recommended by therapeutic guidelines. The decision to use this dichotomy was also based on an earlier analysis, which found that picking up less than 75% of prescriptions during the first year of follow-up was predictive of shorter survival .
The primary analysis was restricted to HIV-infected men and women who were antiretroviral naive and were first prescribed triple drug antiretroviral therapy between 1 August 1996 and 31 July 2000. Study subjects were initially prescribed triple drug combination therapy with regimens including two nucleoside reverse transcriptase inhibitors and either a protease inhibitor (PI) or a non-nucleoside reverse transcriptase inhibitor (NNRTI) at the discretion of the enrolling physician. For the purposes of this analysis we followed the intent-to-treat principle, and subjects were included as they were first dispensed antiretrovirals regardless of whether they later modified their therapeutic regimen.
Cumulative mortality rates were estimated using Kaplan–Meier methods. Event-free subjects were right censored as of 31 July 2001. Participants were pooled into three baseline CD4 cell count groups based upon breakpoints identified previously . These groups were < 50, 50–199, and ≥ 200 × 106 cells/l, respectively. We then further stratified the three CD4 cell count groupings into those with HIV-experienced and non-experienced physicians and into those who were adherent and non-adherent as described above.
Cox proportional hazards regression was used to calculate univariate and adjusted relative hazards and 95% confidence intervals (CIs) . The assumption of proportional hazards was validated by inspection of log [−log (survival function)] estimates against log time plots. Variables examined in this analysis included: baseline plasma HIV-1 RNA levels (log10 transformed); baseline CD4 cell count (< 50, 50–199, and ≥ 200 × 106 cells/l); protease-inhibitor use in the initial regimen (yes versus no); time of initiation of therapy (after versus before July 1997); a prior diagnosis of AIDS (yes versus no); age (continuous); sex; and physician experience and adherence as described above. We also conducted a sub-analysis of those participants that initiated therapy after July 1997, in order to rule out the possibility of residual confounding, that may not be eliminated by simple adjustment with this temporal cut-off. All statistical analyses were performed using SAS software version 6.0 (SAS, Cary, North Carolina, USA).
During the study period, 1580 antiretroviral-naive participants began triple combination therapy with a PI or NNRTI. Of these, 164 (10.4%) were excluded from this analysis for not having both baseline CD4 cell count and plasma HIV-1 RNA level measures available within 6 months prior to the start of antiretroviral therapy. Therefore, the study sample was based on 1416 (90.0%) subjects [1194 (84.3%), men; 222 (15.7%) women]. No differences in sex, and AIDS at baseline, and subsequent mortality was observed between the study sample and those excluded. However, persons excluded from this analysis were more likely to be younger (median age 36 versus 37 years; P = 0.04) and taking protease inhibitors (78.7 versus 69.0%; P = 0.01). At baseline, the median CD4 cell count was 270 × 106 cells/l [inter-quartile range (IQR), 130–420] and median plasma HIV RNA level was 119 500 copies/ml (IQR, 38 000–300 000). As of 31 July 2001, 132 patients had died in the study population. Twenty-two of these were not attributed to HIV and were censored as non-events at the time of death. These 22 deaths included five suicides and 17 accidental drug overdoses. The remaining 110 deaths gave a crude AIDS-related mortality rate of 7.8%.
Figure 1 shows the 42-month Kaplan–Meier survival estimates for the overall cohort stratified by CD4 cell count using the three previously identified CD4 cell count strata . The number of patients in each strata are shown on the figures. Statistically significant differences in survival were noted among the three CD4 strata (all log rank P < 0.05).
Figure 2 shows the Kaplan–Meier plots of time to death in the three CD4 cell count groups stratified by physician experience. Among participants whose enrolling primary care physician was experienced (Fig. 2a), statistically significant differences in survival were noted between those with ≥ 200 × 106 cells/l at baseline and the other two CD4 strata (log rank P < 0.05); however, there was no longer a significant difference between the < 50 × 106 cells/l and 50–199 × 106 cells/l CD4 cell count groupings. Among those whose enrolling primary care physician was non-experienced (Fig. 2b), statistically significant differences in survival were noted among the three CD4 strata (all log rank P < 0.05). When comparisons were made across the two physician experience strata, a statistically significant difference was found between the two < 50 × 106 cells/l groupings, with higher mortality found among the patients with less experienced physicians (P = 0.013).
Figure 3 shows the Kaplan–Meier plots of the time to death in the three CD4 cell count groups, stratified according to their level of adherence. Among those who were adherent (Fig. 3a), statistically significant differences in survival were noted between those with ≥ 200 × 106 cells/l at baseline and the other two strata (log rank P < 0.05), however, the difference between the < 50 × 106 cells/l and 50–199 × 106 cells/l CD4 cell count strata was not statistically significant. Among non-adherent patients (Fig. 3b), statistically significant differences in survival were noted between all CD4 cell count strata (all log rank P < 0.05). When comparisons were made across the two adherence strata, significant differences were found between all pairs of CD4 cell count groupings with mortality being higher among non-adherent patients (all log rank P < 0.05). Of note, among adherent patients with baseline CD4 cell counts ≥ 200 × 106 cells/l, the cumulative mortality rate was only 2.6% after a full 42 months of treatment (Fig. 2a), and we found no differences in survival among CD4 strata above 200 (data not shown). The median baseline CD4 cell count was 365 × 106 cells/l (IQR, 270–460) among the 14 (2.6%) deaths among adherent patients with baseline CD4 ≥ 200 × 106 cells/l.
Table 1 shows the analyses of factors associated with the time to death for the overall cohort. In univariate analyses, a clinical diagnosis of AIDS at baseline, older age, physician experience, adherence, and a CD4 cell count at baseline of less than 200 × 106 cells/l and a higher baseline HIV-1 RNA, were associated with mortality. Date of therapy initiation, gender, history of injection drug use, and protease inhibitor use were not statistically significant as predictors of mortality. In the multivariate analyses, in comparison to patients with a baseline CD4 cell count ≥ 200 × 106 cells/l, the relative hazard of death was 3.81 (95% CI, 2.33–6.24), and it was 7.08 (95% CI, 4.08–12.26) for patients with a baseline CD4 cell counts 50–199 × 106 cells/l and < 50 × 106 cells/l, respectively in adjusted analyses. Similarly, the relative hazard of death was 0.67 (95% CI, 0.45–0.99) amongst participants whose physician was experienced, and the relative hazard of death among patients who were adherent was 0.30 (95% CI, 0.20–0.45) in adjusted analyses. These estimates were not significantly affected by adjusting for additional combinations of non-significant variables, although we chose to force the history of injection drug use variable into the model in order to adjust for patient characteristics that may be associated with non-compliance to medical care . No statistically significant interactions were found, and no interactions were entered into the final model, because they did not improve the model's fit.
In confirmatory analyses, we found that results were consistent when we restricted the analysis considering only the 1053 participants who initiated therapy after 1 July 1997. These results are shown in Table 2. We also evaluated whether residual confounding due to the imprecision of our adherence estimate might explain the association between physician experience and survival. When we conducted analyses to evaluate potential associations between physician experience and adherence, we found that adherence was not associated with physician experience [(chi-square) P = 0.128]. In addition, we also repeated the survival analyses shown in Table 1 considering all cause mortality (including overdoses, suicides, etc). In the latter analysis, physician experience was not independently associated with survival, whereas adherence remained significantly associated with survival in all models.
Finally, in order to provide adjusted estimates of the risk of death by CD4 cell count and physician experience categories, we repeated the Cox regression analyses with variables representing each CD4 and physician experience strata shown in Table 3. Covariates adjusted for are noted on the tables, and the reference category is those patients with ≥ 200 × 106 CD4 cells/l at baseline who were seen by an experienced physician. Shown here, for patients with less than 50 × 106 cells/l at baseline, the adjusted relative hazard of death for patients with experienced physicians was 5.07 (95% CI, 2.50–10.26) and was 11.99 (95% CI, 6.33–22.74) for patients with non-experienced physicians. We also assumed that residual confounding may persist, despite adjustment for baseline AIDS, if the type and severity of baseline AIDS cases was not evenly distributed between physician experience groups, and repeated the above analysis excluding the 182 patients who had a clinical diagnosis of AIDS prior to initiating treatment. Here, the adjusted relative hazard of death for patients with less than 50 × 106 cells/l was 4.27 (95% CI, 1.75–10.43) for patients with experienced physicians and it was 15.80 (95% CI, 8.08–30.89) for patients with non-experienced physicians.
We also repeated the analysis, in order to provide adjusted estimates of the risk of death for the CD4 cell count and adherence categories as shown in Table 4. Here, the reference category was the group of patients with ≥ 200 × 106 cells/l at baseline who were adherent. For patients with ≥ 200 × 106 cells/l, the relative hazard of death for non-adherent patients was 3.55 (95% CI, 1.72–7.35) in comparison with adherent patients in the same CD4 group. For patients with 50–199 × 106 cells/l, the relative hazard of death for adherent patients was 4.54 (95% CI, 2.37–8.70) and was 9.78 (95% CI, 4.48–21.31) for non-adherent patients. For patients with less than 50 × 106 cells/l, the relative hazard of death for adherent patients was 6.19 (95% CI, 3.03–12.65) and was 35.71 (95% CI, 16.17–78.85) for non-adherent patients.
Our results show that survival rates following the initiation of antiretroviral therapy are dramatically improved among patients starting with CD4 cell counts below 200 × 106 cells/l once adjusted for conservative estimates of physician experience and patient adherence. These data suggest that a CD4 cell count of 200–350 × 106 cells/l does not represent an irreversible biological threshold beyond which response to therapy is compromised. Conversely, inappropriate care of patients with advanced disease and patient non-adherence may be the strongest determinants of patient survival, rather than the time at which antiretroviral therapy is initiated prior to 200 × 106 cells/l. As mortality was elevated among all CD4 strata below 200 × 106 cells/l, however, our data do not support delaying the initiation of HAART below this level, and close follow-up will be necessary to examine if survival differences emerge among patients with baseline CD4 cell counts > 200 × 106 cells/l.
Previous studies have shown that patients are not at immediate risk of death until the CD4 cell count drops below 50 × 106 cells/l [30,31], and that AIDS-related opportunistic illnesses (OIs) are rare after the initiation of HAART, except among patients with less than 50 × 106 cells/l . However, it has also been shown that physicians who have treated less than six HIV-infected patients are significantly less knowledgeable about prescription of prophylaxis against OIs, provision of preventative vaccinations, as well as the treatment of acute OIs . These findings probably explain why physician experience had the largest impact on highly advanced patients, and why physician experience was found to be independent of adherence. Additional evidence that the association between physician experience and survival is attributable to the management of advanced patients and not patient selection factors comes from the fact that physician experience was no longer significant when all-cause mortality was considered. Nevertheless, it was not the objective of the present study to identify the ideal level of experience for the optimal management of HIV-infected individuals, and further study will be required to identify the physician-related factors, such as treatment or prevention of opportunistic infections, that may be responsible for our observations. Although there has been debate about the role of physician experience in the HAART era [14, 34–38], to our knowledge, these results are the first to demonstrate an independent effect of physician experience on survival since the widespread use of triple therapy.
Our results are also the first to demonstrate statistically significant differences in survival among adherent and non-adherent patients at all levels of CD4 count, and in particular the dramatic impact adherence has on the survival of highly advanced patients. Although the multiple stratifications shown in Table 4 resulted in moderately large confidence intervals, these findings will contribute to the debate over when therapy is best initiated , and will be useful for guidelines considering the management of the large proportion of patients that continue to present for treatment when CD4 cell counts have decreased to below 200 × 106 cells/l [3, 39].
Previous studies have demonstrated that measuring daily adherence to antiretroviral therapy can be fraught with difficulties that may over or underestimate a patient's actual exposure to treatment [40–43]. Although using refill compliance as a surrogate for adherence has been previously validated [23,27], there is likely a very strong conservative bias operating in our study because patients may have been less than optimally adherent to daily treatment despite receiving ≥ 75% medication during the first year of therapy. It has consistently been found that very high daily adherence is required, to achieve an undetectable plasma viral load, to prevent viral rebound, and to prevent the evolution of resistant virus [17,24, 44,45]. Further study will be required to determine the optimal level of adherence which, as illustrated above, is likely to vary depending on the nature of the endpoint selected (virological, immunological, or clinical) as well as the duration of follow-up. In addition, as we have previously discussed , a limitation of the present study is that patients were defined as adherent on the basis of their behavior during the first year of therapy and then assigned to adherent strata for an analysis of baseline characteristics. Although we adjusted our analyses for pertinent demographic and clinical characteristics, as with all studies of patients treated in observational cohorts, unmeasured differences may exist among study populations and for this reason caution is warranted. However, it was not the objective of the present study to elucidate the level of adherence needed to optimize the outcomes associated with antiretroviral therapy.
In summary, we found that mortality was dramatically altered by very conservative adjustments for physician experience and adherence, suggesting that inappropriate care of advanced patients and patient non-adherence may be the strongest determinants of patient survival, rather than the time at which therapy is initiated prior to CD4 cell counts of 200 × 106 cells/l. As mortality was elevated among all CD4 strata below 200 × 106 cells/l, our data do not support delaying the initiation of HAART to any level below 200 × 106 cells/l. These results should be useful for the development of therapeutic guidelines, and should be helpful to emphasize the role of physician experience and patient adherence in the treatment of HIV infection.
We thank Bonnie Devlin, Diane Campbell, Elizabeth Ferris, Nada Gataric, Kelly Hsu Myrna Reginaldo, Peter Vann for their research and administrative assistance. Particular thanks goes to Keith Chan, Kevin Craib, Mark Tyndall, and Martin Schechter for their advice on the statistical methods.
Sponsorship: E.W. is supported by the canadian Institutes for Health Research and the British Columbia Health Research Foundation. This work was supported by the Michael Smith Foundation for Health Research through a Career Investigator Award and by Canadian Institutes of Health Research through an Investigator Award to R.H.
1.Hammer SM, Squires KE, Hughes MD, Grimes JM, Demeter LM, Currier JS, et al
. A controlled trial of two nucleoside analogues plus indinavir in persons with human immunodeficiency virus infection and CD4 cell counts of 200 per cubic millimeter or less. AIDS Clinical Trials Group 320 Study Team. N Engl J Med
2.Montaner JS, Reiss P, Cooper D, Vella S, Harris M, Conway B, et al
. A randomized, double-blind trial comparing combinations of nevirapine, didanosine, and zidovudine for HIV-infected patients: the INCAS Trial. Italy, The Netherlands, Canada and Australia Study. JAMA
3.Hogg RS, Yip B, Chan KJ, Wood E, Craib KJ, O'Shaughnessy MV, et al
. Rates of disease progression by baseline CD4 cell count and viral load after initiating triple-drug therapy. JAMA
3.Chen R, Westfall A, Coud G, Chatham A, Acosta E, Raper J, et al
. Long-term survival after initiation of antiretroviral therapy. Eighth Conference on Retroviruses and Opportunistic Infections
, Chicago, February 2001 [abstract 341].
5.Sterling TR, Chaisson RE, Moore RD. CD4 T-lymphocytes, and clinical response to highly active antiretroviral therapy. AIDS
, 2001, 15
5.Kaplan J, Hanson D, Karon J, Cohn D, Thompson M, Buskin S, et al
. Late initiation of antiretroviral therapy (at CD4+ lymphocytes < 200 Cells/mL) is associated with increased risk of death. Eighth Conference on Retroviruses and Opportunistic Infections
, Chicago, February 2001 [abstract 520].
7.Egger M, May M, Chene G, Phillips AN, Ledergerber B, Dabis F, et al
. Prognosis of HIV-1-infected patients starting highly active antiretroviral therapy: a collaborative analysis of prospective studies. Lancet
7.DHHS Panel on clinical practices for treatment of HIV infection. Guidelines for the use of Antiretroviral Agents in HIV-Infected Adults and Adolescents.
Updated February 4, 2002. Available at: http://www.hivatis.org
. Accessed May 8, 2002.
9.Pomerantz RJ. Initiating antiretroviral therapy during HIV infection: confusion and clarity. JAMA
10.British HIV Association. British HIV Association (BHIVA) guidelines for the treatment of HIV-infected adults with antiretroviral therapy. HIV Med
11.Yeni PG, Hammer SM, Carpenter CC, Cooper DA, Fischl MA, Gatell JM, et al
. Antiretroviral treatment for adult HIV infection in 2002: updated recommendations of the International AIDS Society–USA Panel. JAMA
12.Kitahata MM, Koepsell TD, Deyo RA, Maxwell CL, Dodge WT, Wagner EH. Physicians’ experience with the acquired immunodeficiency syndrome as a factor in patients’ survival. N Engl J Med
13.Montaner JS, Mellors JW. Antiretroviral therapy for previously treated patients. N Engl J Med
14.Lewis CE. Management of patients with HIV/AIDS. Who should care? JAMA
15.Gallant JE. Strategies for long-term success in the treatment of HIV infection. JAMA
16.Paterson DL, Swindells S, Mohr J, Brester M, Vergis EN, Squier C, et al
. Adherence to protease inhibitor therapy and outcomes in patients with HIV infection. Ann Intern Med
17.Gross R, Bilker WB, Friedman HM, Strom BL. Effect of adherence to newly initiated antiretroviral therapy on plasma viral load. AIDS
18.Haubrich RH, Little SJ, Currier JS, Forthal DN, Kemper CA, Beall GN, et al
. The value of patient-reported adherence to antiretroviral therapy in predicting virologic and immunologic response. California Collaborative Treatment Group. AIDS
19.Hogg RS, Heath KV, Yip B, Craib KJ, O'Shaughnessy MV, Schechter MT, et al
. Improved survival among HIV-infected individuals following initiation of antiretroviral therapy. JAMA
20.Carpenter CC, Fischl MA, Hammer SM, Hirsch MS, Jacobsen DM, Katzenstein DA, et al
. Antiretroviral therapy for HIV infection in 1996. Recommendations of an international panel. International AIDS Society–USA. JAMA
21.Carpenter CC, Fischl MA, Hammer SM, Hirsch MS, Jacobsen DM, Katzenstein DA, et al
. Antiretroviral therapy for HIV infection in 1997. Updated recommendations of the International AIDS Society–USA panel. JAMA
22.World Health Organization. Acquired immune deficiency syndrome (AIDS): interim proposal for a WHO staging system for HIV infection and disease. Wkly Epidemol Rec
23.Steiner JF, Prochazka AV. The assessment of refill compliance using pharmacy records: methods, validity, and applications. J Clin Epidemiol
24.Low-Beer S, Yip B, O'Shaughnessy MV, Hogg RS, Montaner JS. Adherence to triple therapy and viral load response. J Acquir Immune Defic Syndr
25.Palepu A, Yip B, Miller C, Strathdee S, O'Shaughnessy MV, Montaner JS, et al
. Factors associated with the response to antiretroviral therapy among HIV-infected patients with and without a history of injection drug use. AIDS
26.Wood E, Montaner JS, Chan K, Tyndall MW, Schechter MT, Bangsberg D, et al
. Socioeconomic status, access to triple therpay, and survival from HIV-disease since 1996. AIDS
27.Hogg RS, Heath KV, Bangsberg D, Yip B, Press N, O'Shaughnessy MV, et al
. Intermittent use of triple combination therapy is predictive of mortality at baseline and after one year of follow-up. AIDS
28.Cox DR. Regression models and life tables. J Royal Stat Soc B
29.Anis AH, Sun H, Guh DP, Palepu A, Schechter MT, O'Shaughnessy MV. Leaving hospital against medical advice among HIV-positive patients. CMAJ
30.Phillips AN, Elford J, Sabin C, Bofill M, Janossy G, Lee CA. Immunodeficiency and the risk of death in HIV infection. JAMA
31.Ehmann WC, Eyster ME, Wilson SE, Andes WA, Goedert JJ. Relationship of CD4 lymphocyte counts to survival in a cohort of hemophiliacs infected with HIV. Multicenter Hemophilia Cohort Study. J Acquir Immune Defic Syndr
32.Ledergerber B, Egger M, Erard V, Weber R, Hirschel B, Furrer H, et al
. AIDS-related opportunistic illnesses occurring after initiation of potent antiretroviral therapy: the Swiss HIV Cohort Study. JAMA
33.Montaner JS, Hogg RS, Heath KV, Phillips P, Craib KJ, Schechter MT, et al
. Heterogeneity of physician agreement with recommended therapeutic guidelines for the management of HIV-associated disease. Antivir Ther
34.Zuger A, Sharp VL. 'HIV specialists': the time has come. JAMA
35.Dooha S. Controversies: the role of HIV specialists. JAMA
:833; discussion 835.
36.Nitta AT. Controversies: the role of HIV specialists. JAMA
:834; discussion 835.
37.Legg JJ, Balano KB, Goldschmidt RH. Controversies: the role of HIV specialists. JAMA
:833–834; discussion 835.
38.Strelnick AH, Futterman D, Carrascal A, Gillette PN, Murayama R, Weiss C, et al
. Controversies: the role of HIV specialists. JAMA
39.Gupta SB, Gilbert RL, Brady AR, Livingstone SJ, Evans BG. CD4 cell counts in adults with newly diagnosed HIV infection: results of surveillance in England and Wales, 1990
–1998. CD4 Surveillance Scheme Advisory Group. AIDS
40.Liu H, Golin CE, Miller LG, Hays RD, Beck CK, Sanandaji S, et al
. A comparison study of multiple measures of adherence to HIV protease inhibitors. Ann Intern Med
41.Duran S, Solas C, Spire B, Carrieri MP, Fuzibet JG, Costagliola D, et al
. 'Do HIV-infected injecting drug users over-report adherence to highly active antiretroviral therapy?’ A comparison between patients’ self-reports and serum protease inhibitor concentrations in the French Manif 2000 cohort study. AIDS
42.Bangsberg DR, Hecht FM, Clague H, Charlebois ED, Ciccarone D, Chesney M, et al
. Provider assessment of adherence to HIV antiretroviral therapy. J Acquir Immune Defic Syndr
43.Wendel CS, Mohler MJ, Kroesen K, Ampel NM, Gifford AL, Coons SJ. Barriers to use of electronic adherence monitoring in an HIV clinic. Ann Pharmacother
44.Gallego O, de Mendoza C, Perez-Elias MJ, Guardiola JM, Pedreira J, Dalmau D, et al
. Drug resistance in patients experiencing early virological failure under a triple combination including indinavir. AIDS
45.Le Moing V, Chene G, Carrieri MP, Alioum A, Brun-Vezinet F, Piroth L, et al
. Predictors of virological rebound in HIV-1-infected patients initiating a protease inhibitor-containing regimen. AIDS