1543. The Effect of Antiretroviral Therapy Use Among HIV-1 Positive Children on the Hazard of AIDS Using Calendar Year as an Instrumental Variable
Session: Poster Abstract Session: HIV Pediatric and Adolescents
Friday, October 28, 2016
Room: Poster Hall
Posters
  • ID.poster.IDSA.2016.v4.pdf (1.8 MB)
  • Background:

    Instrumental variable (IV) analyses are now a common causal inference technique used in the absence of randomized data. Antiretroviral Therapy (ART) was first introduced in 1996. Therefore, calendar periods have been used as a proxy for ART use. However, ART use misclassification can bias IV analyses.

    Methods:

    The authors performed an adapted IV analysis of 267 children perinatally-infected with human immunodeficiency virus (HIV-1) followed from 1988 to 2009 to estimate the effect of ART on the hazard of progression to category C diagnosis. Noncompliance adjustments were used to correct for misclassification of ART use.

    Results:

    During 48,380 person-days, 106 HIV-positive children progressed to C diagnosis or death. The intention to treat (ITT) rate ratio of C diagnosis or death comparing the pre-ART and ART eras was estimated at 2.62 (95% confidence interval (CI): 1.58, 4.35). An IV estimator was used to adjust for ART use misclassification, yielding an IV rate ratio of 3.93 (95% CI: 2.37, 6.53). Weighting by the inverse probability of calendar era, given covariates, did not significantly alter the results.

    Conclusion:

    ART use decreases progression to C diagnosis or death. The use of noncompliance adjustments for ART misclassification in IV analyses may provide more robust evidence of ART's effectiveness than traditional ITT analysis.

    Distribution of Events (First Category C Diagnosis or Death) and Person-Days by Calendar Period Among Children With Previous Category B Diagnosis

    Calendar Period

    (Before 1998/ After 1998)

    No. of events First C Diagnosis or Death

    No. of Person-Days

    Ratea

    Intent To Treat

    Instrumental Variable (IV)

    Rate Difference

    95% CI

    Rate Ratio

    95% CI

    Rate Ratio

    95% CI

    Unweightedb

    Pre-ART

    88

    31,519

    2.79

    1.72

    0.96,

    2.48

    2.62

    1.58, 4.35

    3.93

    2.37,

    6.53

    ART

    18

    16,861

    1.07

    0

    1

    1

    Total

    106

    48,380

    2.19

    Weightedc,d

    Pre-ART

    86.2

    31,376

    2.75

    1.20

    -1.04,

    2.78

    1.78

    0.71,

    4.41

    2.46

    0.93,

    10.33

    ART

    21.8

    14,113

    1.55

    0

    1

    1

    Total

    108.0

    45,489

    2.37

    Abbreviations: ART (antiretroviral therapy)

    a Events per 1,000 person-days

    b Confidence intervals (CI) for unweighted intent-to-treat (ITT) estimates were calculated by using standard formulas

    c CI for weighted ITT estimates and IV estimates were calculated by bootstrap

    d Covariates included race/ethnicity.

    Andrew Anglemyer, PhD, MPH, Operations Research, Naval Postgraduate School, Monterey, CA, Amy Sturt, MD, Infectious Diseases and Geographical Medicine, Stanford University School of Medicine, Stanford, CA and Yvonne Maldonao, MD, Pediatric Infectious Diseases, Stanford University School of Medicine, Stanford, CA

    Disclosures:

    A. Anglemyer, None

    A. Sturt, None

    Y. Maldonao, None

    Findings in the abstracts are embargoed until 12:01 a.m. CDT, Wednesday Oct. 26th with the exception of research findings presented at the IDWeek press conferences.