Presentation to HIV care and antiretroviral therapy initiation and response in clinical practice from 2003 through 2013

Abstract

Universal HIV screening and treatment initiation of HIV-positive persons are well-established standards. However, late presentation to care is a barrier to early antiretroviral therapy (ART) and prevention of HIV transmission. We sought to determine the immunodeficiency at presentation to care and characterize the initiation and response to ART among HIV-positive persons over 2003–2013 in our urban HIV clinical practice at the Cleveland Clinic. Using a retrospective cohort study design, we assessed the CD4 cell count of HIV-positive patients at entry into care for each year and evaluated the trend over time. For patients who initiated treatment, we assessed the pretreatment CD4 cell count, consistency of timing and regimen with US treatment guidelines, and HIV RNA level at one-year and last follow-up visits. Regression analyses were used to determine predictors of study outcomes. We found that the cohort (N = 452) median CD4 cell count at presentation to care was 297 cells/mm³ (inter-quartile range: 104–479 cells/mm³), without any significant change over time (P = 0.62), and with 37% and 21% of presentations being late and advanced, respectively. Guideline-consistency (85%–100%) and regimen-consistency (41%–100%) were moderate to high and improved over time. Virologic suppression (<400 copies/ml) at one year and last follow-up was high (79% and 92%) and associated with regimen selection and durability. We conclude that CD4 cell count at first presentation to HIV care remained less than 350 cells/mm³ for 11 years in our clinical practice, despite advances in HIV testing and treatment guidelines. Early diagnosis and linkage to care and treatment are critical for ending the HIV epidemic.

Keywords

HIV antiretroviral therapy treatment AIDS

Introduction

Over a million persons are living with HIV in the United States, with an estimated 15% unaware of their infection.¹ These unaware persons account for approximately 40% of the 40,000 annual new infections.¹ While advances in HIV prevention and care have brought the goal of ending the epidemic within our reach,² ongoing HIV transmission is a major hurdle. Given its remarkable efficacy for treatment and prevention, antiretroviral therapy (ART) has become the cornerstone of global³ and national⁴ HIV response strategies and the HIV treatment guidelines^5,6 have swung back from conservative to progressive with treatment now recommended for all HIV-positive persons. Furthermore, the gains in antiretroviral potency, tolerability and simplicity, facilitate improved virological, immunological, clinical, and epidemiological outcomes.^5,6

Expanded HIV testing recommendations⁷ raise expectations for wider and frequent testing alongside earlier linkage and presentation to care. However, late entry to HIV care is common around the globe.^8–14 Most of the United States studies in the modern ART era (through 2010) have examined diverse data^8,10 and demonstrated no meaningful increase in CD4 cell count at presentation to care over time. We sought to evaluate the time-trends in presentation to care among HIV-positive patients over 11 years (2003–2013) paired with an assessment of the consistency of antiretroviral initiation and selection with the Department of Health and Human Services (DHHS) guidelines⁵ at time of treatment initiation, and determined the extent and durability of virological and immunological response to ART.

Methods

Study design and population

A retrospective cohort study design was used to analyze routinely collected data at the Cleveland Clinic Main Campus outpatient infectious disease clinic that provides care for approximately 700 HIV-infected patients annually in an urban setting and offers Ryan White Grant funded services. HIV-positive patients aged 18 years and older with no history of antiretroviral exposure, who presented to care between 1 January 2003 and 31 December 2013, were eligible for inclusion in the study. Patients were excluded if they were treatment-experienced or did not have a CD4 cell count measured at their initial visit. A subset population included patients who initiated ART at the Cleveland Clinic during the study period and had at least one year of follow-up, with presentation during or before the study period.

Study data

The following data were collected for the study population using the electronic health record (Epic). Demographic and epidemiological data included date of birth, gender, race, and HIV transmission risk category (men who have sex with men [MSM], heterosexual contact, injection drug use, transfusion-related). Immunological and virological data comprised the CD4 cell count and plasma HIV RNA levels at presentation to care, initiation of ART, one-year follow-up, and last follow-up. Antiretroviral treatment-related data comprised the initial antiretroviral regimen, change in regimen and regimen at last follow-up, duration in care, and the DHHS treatment guidelines in effect at time of treatment initiation.

The primary endpoint was the CD4 cell count at presentation to care over time. Secondary endpoints included: (i) guideline-consistency of treatment initiation, (ii) regimen-consistency of antiretroviral prescription, (iii) change in treatment regimen, (iv) change in CD4 cell count, (v) virological suppression at one-year follow-up, (vi) virological suppression at last follow-up, and (vii) determinants of virological response.

Variable definitions

Presentation to care was defined as the first clinic visit for HIV care with a measured CD4 cell count. Late presentation to care was defined as CD4 cell count >200 but <350 cells/mm³ while advanced presentation was CD4 cell count <200 cell/mm³ and/or clinical acquired immunodeficiency syndrome (AIDS).^15,16

We defined one-year follow-up as the first evaluable clinic and/or laboratory visit, between 9 and 15 months after initiation of ART. Last follow-up was the last evaluable clinic and/or laboratory visit, at least three months after the one-year follow-up until 31 December 2013.

Timing of antiretroviral treatment and regimens at initiation were evaluated for concordance with the DHHS guidelines in effect at the initiation date; guideline-consistency was defined as initiation of treatment at the recommended CD4 cell count, presence of an AIDS-defining illness, and/or plasma HIV RNA level. Regimen-consistency was the initiation of a ‘recommended or preferred’ first-line antiretroviral regimen (initial regimen).

Virological suppression was primarily defined as plasma HIV RNA level <400 copies/ml. Though levels <20–50 copies/ml were also evaluated, a threshold of <400 copies/ml was chosen to avoid confounding due to false low level viremia.^17,18 We defined antiretroviral regimen switch as a change in therapy from the initial regimen at one-year and/or last follow-up. Changes in antiretroviral formulation from multiple tablets to a single dose regimen did not constitute a regimen switch.

Statistical analysis

Summary statistics (n, mean, median, inter-quartile range [IQR], standard deviation for continuous variables, and frequency counts for categorical variables) for outcomes were generated per calendar year and overall for two patient groupings: All patients who presented to care between 2003 and 2013 and all patients who were initiated on therapy between 2003 and 2013 and had one-year follow-up measurements excluding those who had died or were lost to follow-up. Univariate and multivariate correlations between outcome and predictor variables were developed using logistic and mixed model regression methods.¹⁹ Predictor variables were tested for independence from one another using condition indices. Variables with condition indices ≤10 were considered to be sufficiently independent of one another to be used in the multivariable model. Backward elimination regression methods were used to generate reduced models (models containing only statistically significant terms [P < .05]) for each outcome of interest. Analyses were conducted with SAS, version 9.4 (SAS Institute, Cary NC, USA).

Ethical approval

The research protocol was approved by the institutional review board at Cleveland Clinic.

Results

Population characteristics

A total of 452 patients met the eligibility criteria for presentation to care during 2003–2013 and were included in the study, of whom 406 initiated ART during 2003–2013, while 46 were lost to follow-up. An additional 24 patients initiated treatment during 2003–2013, most of whom (22) presented before 2003. Of the total 430 patients initiating ART, 314 were evaluable at one year. The presenting population was predominantly male (85%), MSM (52%), and African American (49%) or White (47%), with a median age of 37 years (IQR: 27–45 years).

Timing of presentation to care and treatment initiation

The cohort’s median CD4 cell count at the time of presentation to HIV care was 297 cells/mm³ (IQR: 104–479 cells/mm³; N = 452). A multivariable regression analysis of presentation CD4 cell count as a function of CD4 stratum (<200, 200–350, >350 cells/mm³), calendar year, gender and patient age, indicated no significant change over time (P = 0.62) (Figure 1). Overall, 37% of patients presented with a CD4 cell count between 200 and 350 cells/mm³, 21% presented with a CD4 cell count <200 cells/mm³, and 53% were advanced presenters with CD4 cell count <200 cells/mm³ and/or clinical AIDS; these proportions also remained stable during the study period. Eight patients (1.7%) presented with acute HIV infection and the observed spectrum of AIDS-defining illnesses included: candidiasis (n = 47); pneumocystis pneumonia (n = 29); HIV wasting (n = 17); lymphoma (n = 13); neurologic manifestations including HIV associated dementia and HIV encephalopathy (n = 9); cytomegalovirus infection (n = 8); recurrent Staphylococcal infection and other dermatologic manifestations (n = 7); herpes simplex virus infections and Kaposi’s sarcoma (n = 5 each); cryptococcal meningitis; toxoplasmosis infection; Mycobacterium avium-intracellulare infection and herpes zoster infection (n = 4 each); progressive multifocal leukoencephalopathy and HIV associated nephropathy (n = 3 each); cervical cancer and oral leukoplakia (n = 2 each); recurrent pneumonia and anal neoplasm (n = 1 each).

By the time of treatment initiation, the CD4 cell count of the initiating cohort fell to a median of 253 cells/mm³ (IQR: 78–404 cells/mm³; N = 430). The cohort’s median HIV RNA at the time of presentation to HIV care was 84,700 copies/ml (IQR: 22,000–295,000 copies/ml; N = 446). By the time of treatment initiation, the median HIV RNA rose to 104,000 copies/ml (IQR: 30,800–294,000 copies/ml; N = 427). Overall, the median time lag between time of presentation and initiation was 35 days. The median lag time per calendar year showed a decreasing trend over the study, ranging from ≥305 days in 2007 to ≥33 days in 2013.

Guideline-consistency and regimen-consistency

Consistency with DHHS guideline criteria for initiation of ART was high overall (range: 85%–100%) and improved over time (Table 1). In multivariate analysis, the odds ratio (OR) of maintaining guideline-consistency for treatment initiation increased by 39% for each calendar year from 2003 to 2013 (OR: 1.39, 95% confidence interval [CI]: 1.18–1.65).

Table 1.

Summary of antiretroviral therapy by year of initiation (N = 430).

Characteristics	2003n = 24	2004n = 31	2005n = 22	2006n = 42	2007n = 49	2008n = 29	2009n = 38	2010n = 43	2011n = 47	2012n = 52	2013n = 53
Guideline-consistency at initiation of ART, n = 430
Initiation criteria^a	96%	90%	91%	86%	94%	93%	87%	100%	100%	100%	100%
ART regimen^b	58%	67%	41%	69%	98%	90%	100%	97%	96%	94%	87%
Regimen type at initiation of ART, n = 430
PI-based + 2 NRTIs	25%	42%	59%	36%	29%	17%	16%	14%	19%	21%	23%
NNRTI-based + 2 NRTIs	67%	58%	36%	64%	71%	80%	84%	81%	74%	77%	56%
Efavirenz + 2 NRTIs	56%	61%	75%	93%	94%	96%	100%	100%	97%	95%	93%
INSTI-based + 2 NRTIs	0%	0%	0%	0%	0%	3%	0%	5%	7%	2%	19%
PI + NNRTI + NRTI	0%	0%	5%	0%	0%	0%	0%	0%	0%	0%	0%
Other	8%	0%	0%	0%	0%	0%	0%	0%	0%	0%	2%
	n = 19	n = 19	n = 19	n = 32	n = 37	n = 26	n = 32	n = 36	n = 35	n = 35	n = 24
Year 1 visit n = 314
HIV RNA <400 copies/ml	37%	43%	74%	75%	94%	92%	84%	92%	74%	80%	92%
No change in ART regimen^c	84%	68%	79%	81%	84%	96%	84%	94%	91%	83%	83%
	n = 18	n = 18	n = 16	n = 29	n = 33	n = 24	n = 27	n = 31	n = 29	n = 24	n = 0
Last evaluable visit, n = 249
HIV RNA < 400 copies/ml	100%	83%	94%	93%	94%	79%	89%	97%	93%	100%	–
No change in ART regimen^c	17%	38%	63%	62%	79%	75%	78%	84%	69%	83%	–

ART: antiretroviral therapy; INSTI: integrase strand transfer inhibitor; NNRTI: nonnucleoside reverse transcriptase inhibitor; NRTI: nucleoside reverse transcriptase inhibitor; PI: protease inhibitor.

^aDefined as initiation of ART at the recommended CD4 cell count, presence of an AIDS-defining illness, and/or HIV RNA level based on the DHHS guidelines published at the initiation date.

^bDefined as initiation of a ‘recommended or preferred’ ART regimen based on the DHHS guidelines published at the initiation date.

^cNo change in ART regimen compared to that at the time of treatment initiation.

The consistency of initial regimen selection with DHHS guideline recommended or preferred first-line ART varied from moderate to high (range: 41%–100%) and improved from early (2003–2006) compared to middle to late (2007–2013) calendar period (Table 1). A reciprocal trend was seen in the prescription of protease inhibitors (PIs), which decreased during 2007–2013. The most commonly prescribed regimen was efavirenz plus two nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs). In multivariate analysis, regimen-consistency was more common in males (OR: 2.73, 95% CI: 1.20–6.21) and for efavirenz-containing regimens (OR 21.17, 95% CI: 8.45–52.80). The odds of maintaining regimen-consistency increased by 26% for each calendar year of initiation from 2003 to 2013 (OR: 1.28, 95% CI: 1.13–1.40).

Treatment outcomes

One-year follow-up

Three-hundred and fourteen patients were evaluated at the one-year follow-up. Patients were excluded from this analysis for reasons including loss to follow-up, death, failure to take the prescribed therapy, and not following up at 9–15 months of treatment at study closure. At this time point (Figure 2), 85% (268/314) of patients remained on the initial regimen and 79% (248/314) of patients were virologically suppressed (HIV RNA < 400 copies/ml). The CD4 cell count change from treatment initiation to one-year follow-up was a median of +219 cells/mm³. In multivariate regression analysis, the patients without a switch in their initial regimen showed a reduction of 0.72 in log HIV RNA relative to patients with a regimen switch. Furthermore, the odds of virological suppression at one year were 2.5 times higher for those without compared to those with regimen switch. There was a significant difference (P = 0.019) between the suppressed and nonsuppressed patients in terms of their initial regimen. However, the most commonly (66%) prescribed regimen in the cohort was efavirenz plus tenofovir disoproxil fumarate plus lamivudine or emtricitabine (EFV+TDF+XTC); 70% of the suppressed patients received this regimen compared to 49% of the unsuppressed. Trends in the antiretroviral regimen prescription are shown in Table 1.

Last follow-up

Of the patients evaluated at the one-year follow-up, 249 patients were evaluated at the last follow-up (Table 1 and Figure 2). Besides loss to follow-up and death, the reason for exclusion included less than three months since one-year follow-up. The median length of care for patients presenting for last follow-up was 4.6 years. The change in CD4 cell count from treatment initiation to last follow-up was a median of +349 cells/mm³. Compared to one-year follow-up, the proportion of patients remaining on the initial regimen declined at last follow-up (85% versus 67%), but a higher proportion were virologically suppressed (79% versus 92%). Suppressed versus nonsuppressed patients continued to show significant difference with regard to their initial regimen (P = 0.035) at this time point; 68% suppressed versus 42% nonsuppressed received EFV+TDF+XTC. In multivariate regression, patients on a PI-based regimen showed an increase of 0.78 in log HIV RNA relative to patients on an nonnucleoside reverse transcriptase inhibitor (NNRTI)-based regimen. Furthermore, virologic suppression at the one-year and last follow-up was correlated; the odds of virological suppression at last follow-up decreased by 25% for each unit increase in log HIV RNA at one-year follow-up (OR: 0.75, 95% CI: 0.58–0.97, P = 0.029).

Discussion

The main finding of this study is that patients newly entering care at our institution, presented late for more than a decade (2003–2013), with an overall median CD4 cell count of 297 cells/mm³ (IQR: 104–479), despite a sea-change in the guidelines for HIV testing and treatment over the same time period. Our study adds new data to the growing body of evidence showing moderate to no improvement in immunodeficiency at presentation to care in the era of potent ART in both the developed and developing countries.^8–14 Collectively, these data indicate a pervasive delay in HIV diagnosis and linkage to care among HIV-positive persons globally over three decades (1992–2013), including the United States,^8,10 where the number of undiagnosed infections has decreased.^20–22 As shown previously,¹⁴ we also observed a lag (median of 35 days) between presentation and treatment initiation with a further decline in CD4 cell count (to a median 253, IQR 78–404 cells/mm³). However, in contrast to another study,²³ our consistency with DHHS guidelines for both treatment initiation (80%–100%) and regimen selection (41%–100%) was high overall and improved over time.

Despite late presentation to care, patients evaluable at one-year and last follow-up showed progressive immunological recovery following treatment initiation (median 219 and 349 cells/mm³, respectively, at one-year and last follow-up) and high rates (79%–92%, respectively, at one-year and last follow-up) of virological suppression (HIV RNA < 400 copies/ml).²⁴ The latter was associated with the initiation and maintenance of an NNRTI-based regimen, confirming the superiority in clinical practice, of the available first-line preferred regimen during the study period; EFV+TDF+XTC.²⁵ The availability of EFV+TDF+ FTC as a (first) single tablet regimen in 2006 led to a subsequent increase not only in the utilization of this regimen but also the consistency (>90%) of its selection with the prevailing guidelines, likely due to enhanced simplicity and convenience.

Our study has several limitations. First, ours is a single center study with a relatively small sample size and predominantly MSM patients. However, our results are in strong agreement with the literature that shows late entry into HIV care,^8–14 despite major efforts during the study period (2003–2013), toward early diagnosis and treatment. Moreover, we analyzed longitudinal individual-level data for our cohort using a comprehensive electronic health record over 11 years which is a major strength. Second, we could not determine the date of HIV infection and/or diagnosis for most of our patients, because they presented with an established infection. Although the date of HIV infection may be estimated based on the level of immunodeficiency,²⁶ the delay from time of infection/diagnosis to care-presentation is not precisely known for our cohort. The estimate of diagnosis delays in the United States remains high despite a decrease (by seven months), from a median of 3.6 years in 2011 to 3 years in 2015 with a quarter of HIV-positive persons diagnosed ≥7 years after infection.¹ Yet the biggest gap in the HIV care continuum is linkage to care with initiation and continuation of ART.²⁷ Regardless, we did not see a significant change in the presenting CD4 cell count during the study. Third, we used self-report to exclude prior antiretroviral exposure which could have led to misclassification in some cases. However, we carefully assessed the virological and immunological outcomes following treatment initiation. Fourth, missing data are an inherent problem in observational studies including ours.²⁸ For transparency, we used the complete case analytic approach, which may have led to bias in our outcomes.²⁹ The underlying reasons (competing risks) for subjects’ noninclusion at the one-year and last follow-up, plus the nature (random/nonrandom) and association of these reasons with outcomes, were not precisely known. Thus, use of alternate approaches^30–32 would not have led to bias-free results.³³ Finally, though we studied our patients for more than a decade that saw increasing use of integrase strand transfer inhibitors, prescription of dolutegravir, a newer and superior drug,³⁴ approved in 2013, was limited during our study period. The same was true for rapid treatment initiation.³⁵ These issues are to be addressed in future work.

In conclusion, our study confirms late presentation to care in the United States and supports calls for new and improved methods for optimizing rates of HIV diagnosis and linkage to care and treatment.

Footnotes

Authors’ contributions

AP (analytic design, collection, organization and interpretation of data, interpretation of results, writing of first manuscript draft and revisions)

SAP (collection of data, review of findings and manuscript)

RKS (analytic design; review of findings and manuscript)

BY-L (generation and interpretation of virological data; manuscript review)

LHC (review of findings and manuscript)

AJT (review of findings and manuscript)

RSB (statistical analysis, review of findings, manuscript writing and review)

ULA (study design, analytic design; interpretation of data, interpretation of results, writing of first manuscript draft and revisions)

Acknowledgments

We gratefully acknowledge assistance from eReasearch at the Cleveland Clinic for assistance with information retrieval.

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: ULA: none; AP: none; SAP: none; RKS: none; BY-L: none; LHC: none; AJT: Gilead Speaker’s Bureau; RSB: none.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Dailey

Hoots

Hall

, et al. Vital signs: human immunodeficiency virus testing and diagnosis delays – United States. MMWR Morb Mortal Wkly Rep 2017; 66: 1300–1306

UNAIDS. On the fast-track to end AIDS: UNAIDS 2016–2021 strategy, http://www.unaids.org/sites/default/files/media_asset/20151027_UNAIDS_PCB37_15_18_EN_rev1.pdf (2015, accessed 27 December 2018).

UNAIDS. 90–90–90 – an ambitious treatment target to help end the AIDS epidemic, http://www.unaids.org/en/resources/documents/2014/90-90-90 (2014, accessed 27 December 2018).

Fauci AS, Redfield RR, Sigounas G, Weahkee MD, Giroir BP. Ending the HIV Epidemic: A Plan for the United States. JAMA. 2019 Feb 7. doi: 10.1001/ JAMA .2019.1343. [Epub ahead of print] PubMed PMID: 30730529.

Department of Health and Human Services. Archived adult and adolescent guidelines, https://aidsinfo.nih.gov/guidelines/archive/adult-and-adolescent-guidelines/ (accessed 27 December 2018).

World Health Organization. Guidelines and policy briefs on HIV, https://www.who.int/hiv/pub/guidelines/en/ (accessed 27 December 2018).

Branson

Handsfield

Lampe

, et al. Revised recommendations for HIV testing of adults, adolescents, and pregnant women in health-care settings. MMWR Recomm Rep 2006; 55: 1–17.

Althoff

Gange

Klein

, et al. Late presentation for human immunodeficiency virus care in the United States and Canada. Clin Infect Dis 2010; 50: 1512–1520.

Ford

Mills

Egger

Editorial commentary: immunodeficiency at start of antiretroviral therapy: the persistent problem of late presentation to care. Clin Infect Dis 2015; 60: 1128–1130.

10.

Lesko

Cole

Zinski

, et al. A systematic review and meta-regression of temporal trends in adult CD4(+) cell count at presentation to HIV care, 1992–2011. Clin Infect Dis 2013; 57: 1027–1037.

11.

Lin

Cheng

, et al. Trends and outcomes of late initiation of combination antiretroviral therapy driven by late presentation among HIV-positive Taiwanese patients in the era of treatment scale-up. PLoS One 2017; 12: e0179870.

12.

Mocroft

Lundgren

Antinori

, et al. Late presentation for HIV care across Europe: update from the collaboration of observational HIV epidemiological research Europe (COHERE) study, 2010 to 2013. Euro Surveill 2015; 20: 1–12.

13.

Raffetti

Postorino

Castelli

, et al. The risk of late or advanced presentation of HIV infected patients is still high, associated factors evolve but impact on overall mortality is vanishing over calendar years: results from the Italian MASTER cohort. BMC Public Health 2016; 16: 878.

14.

Siedner

Bassett

, et al. Trends in CD4 count at presentation to care and treatment initiation in sub-Saharan Africa, 2002–2013: a meta-analysis. Clin Infect Dis 2015; 60: 1120–1127.

15.

Antinori

Johnson

Moreno

, et al. Report of a European Working Group on late presentation with HIV infection: recommendations and regional variation. Antivir Ther 2010; 15: 31–35.

16.

Sabin

Schwenk

Johnson

, et al. Late diagnosis in the HAART era: proposed common definitions and associations with mortality. AIDS 2010; 24: 723–727.

17.

Briggs

Templeton

Fernando

Comparing HIV viral load assays and frequency of low level virological rebound in clinical practice. Int J STD AIDS 2014; 25: 1029–1034.

18.

Mortier

Vancoillie

Dauwe

, et al. Meticulous plasma isolation is essential to avoid false low-level viraemia in Roche Cobas HIV-1 viral load assays. Antivir Ther 2018; 23: 277–281.

19.

Belsley

Kuh

Welsch

RE.

Regression diagnostics: identifying influential data and sources of collinearity. New York: John Wiley & Sons, 1980.

20.

Vital signs: HIV testing and diagnosis among adults–United States, 2001–2009. MMWR Morb Mortal Wkly Rep 2010; 59: 1550–1555.

21.

Hall

Tang

, et al. Prevalence of diagnosed and undiagnosed HIV infection – United States, 2008–2012. MMWR Morb Mortal Wkly Rep 2015; 64: 657–662.

22.

Satcher Johnson

Song

Hall

. Estimated HIV incidence, prevalence, and undiagnosed infections in US States and Washington, DC, 2010–2014. J Acquir Immune Defic Syndr 2017; 76: 116–122.

23.

Cocohoba

Wang

Cox

, et al. Consistency of initial antiretroviral therapy with HIV treatment guidelines in a US cohort of HIV-infected women. J Acquir Immune Defic Syndr 2008; 47: 377–383.

24.

Phillips

Staszewski

Weber

, et al. HIV viral load response to antiretroviral therapy according to the baseline CD4 cell count and viral load. JAMA 2001; 286: 2560–2567.

25.

Riddler

Haubrich

DiRienzo

, et al. Class-sparing regimens for initial treatment of HIV-1 infection. N Engl J Med 2008; 358: 2095–2106.

26.

Lodi

Phillips

Touloumi

, et al. Time from human immunodeficiency virus seroconversion to reaching CD4+ cell count thresholds <200, <350, and <500 Cells/mm(3): assessment of need following changes in treatment guidelines. Clin Infect Dis 2011; 53: 817–825.

27.

Frieden

Foti

Mermin

Applying public health principles to the HIV epidemic–how are we doing?

N Engl J Med 2015; 373: 2281–2287.

28.

von Elm

Altman

Egger

, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet 2007; 370: 1453–1457.

29.

Karahalios

Baglietto

Carlin

, et al. A review of the reporting and handling of missing data in cohort studies with repeated assessment of exposure measures. BMC Med Res Methodol 2012; 12: 96.

30.

Fine

Gray

RJ.

A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999; 94: 496–509.

31.

Rotnitzky

Robins

Analysis of semi-parametric regression models with non-ignorable non-response. Stat Med 1997; 16: 81–102.

32.

Sterne

White

Carlin

, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ 2009; 338: b2393.

33.

Lau

Lesko

Missingness in the setting of competing risks: from missing values to missing potential outcomes. Curr Epidemiol Rep 2018; 5: 153–159.

34.

Kanters

Vitoria

Doherty

, et al. Comparative efficacy and safety of first-line antiretroviral therapy for the treatment of HIV infection: a systematic review and network meta-analysis. Lancet HIV 2016; 3: e510–e520.

35.

Koenig

Dorvil

Devieux

, et al. Same-day HIV testing with initiation of antiretroviral therapy versus standard care for persons living with HIV: a randomized unblinded trial. PLoS Med 2017; 14: e1002357.