An Early Infant HIV Risk Score forTargeted HIV Testing at BirthNicolette M. Du Plessis, PhD,a Chris J.B. Muller, PhD,b Theunis Avenant, MMed Paed,a Michael S. Pepper, PhD,c

Ameena E. Goga, PhDa,d

abstractBACKGROUND: Early HIV testing is needed for treatment success in young infants, but universaltesting is expensive. In this study, we examined the feasibility of early infant HIV risk scoresfor targeted polymerase chain reaction (PCR) testing and early HIV diagnosis.

METHODS: A cross-sectional cohort of newborns exposed to HIV was enrolled and PCR testedwithin 72 hours. We quantified associations between HIV infection and clinical and laboratorymaternal-infant parameters by logistic regression models and determined sensitivity andspecificity for derived risk scores.

RESULTS: From August 2014 to December 2016, 1759 participants were enrolled. Motherswithout antenatal care (5.7% [97 of 1688]) were more likely to deliver newborns who arePCR-positive (P = .0005). A total of 1 in 5 mothers (217 of 990; 21.9%) had HIV viral load (VL).1000 copies per µL. A total of 432 of 1655 (26.1%) infants were preterm. Low birth weightwas documented in 398 of 1598 (24.55%) and 13 of 31 (40.63%) newborns who are PCR-negative and -positive, respectively (P = .0329). A total of 204 of 1689 (12.08%) were growthrestricted or small for gestational age, and 6 of 37 (16.22%) were PCR-positive. Symptomaticnewborns frequently tested positive (P = .0042). The HIV PCR positivity rate was 2.2% (37 of1703). Two-risk (combined 3-drug antiretroviral therapy [cART] duration, VL), 3-risk (cARTduration, VL, symptomatic newborn), and 4-risk (cART duration, VL, symptomatic, small forgestational age newborn) models for HIV acquisition had predictive probability of 0.28, 0.498,and 0.57, respectively; this could guide targeted birth testing. However, using the 3- and 4-riskscores (probability 0.02 and 0.04), 20% and 24% will be missed compared with universaltesting.

CONCLUSIONS: Targeted newborn testing requires access to maternal VL. Even if risk modelsinclude parameters such as maternal cART history, birth weight, weeks’ gestation, andsymptoms, 1 in 5 newborns who are infected will not be targeted. At present, we supportuniversal PCR testing at birth within the South African prevention of mother-to-childtransmission of HIV context.

WHAT’S KNOWN ON THIS SUBJECT: Early HIV testing at or shortlyafter birth is needed to guarantee timely HIV treatment successfor very young infants, but universal HIV polymerase chainreaction testing is expensive.

WHAT THIS STUDY ADDS: Models to identify infants at risk forintrauterine HIV infection were developed. These findings couldguide a targeted approach to birth HIV testing. Targetedpolymerase chain reaction testing to diagnose HIV infection invery young infants requires access to maternal viral load testing.

To cite: Du Plessis NM, Muller CJB, Avenant T, et al. AnEarly Infant HIV Risk Score for Targeted HIV Testing atBirth. Pediatrics. 2019;143(6):e20183834

aDepartment of Paediatrics, Faculty of Health Sciences and cDepartment of Immunology, SAMRC Extramural Unitfor Stem Cell Research and Therapy, Institute for Cellular and Molecular Medicine, University of Pretoria, Pretoria,South Africa; bDepartment of Statistics and Actuarial Science, University of Stellenbosch, Stellenbosch, SouthAfrica; and dHealth Systems Research Unit, South African Medical Research Council, Pretoria, South Africa

Prof Du Plessis conceptualized and designed the methodology, developed the software for datacollection, collected data, conducted the formal analysis, drafted the original manuscript, andreviewed and edited the final manuscript; Dr Muller conducted the formal analysis, assisted withdata curation, drafted the original manuscript, and reviewed and revised the final manuscript;Profs Avenant and Pepper conceptualized the study, designed the methodology, critically reviewedthe manuscript for important intellectual content, and edited the (Continued)

PEDIATRICS Volume 143, number 6, June 2019:e20183834 ARTICLE by guest on June 22, 2019www.aappublications.org/newsDownloaded from

HIV can be vertically transmittedfrom mother to child antepartum,during labor and delivery, orpostpartum via breastfeeding.1

Despite the success of preventionof mother-to-child transmissionof HIV (PMTCT) programs inreducing intrapartum HIVinfections2,3 in South Africa,proportionately more infants arebeing born with HIV due to in uterotransmission.4

Early infant diagnosis (EID)and treatment of HIV remainsa global public health priority.During the Children with HIV EarlyAntiretroviral Therapy trial, infantswere randomly assigned to receivedeferred therapy versus earlycombined 3-drug antiretroviraltherapy (cART) at a median ageof 7.4 weeks.5 Early HIV diagnosisand antiretroviral therapy,irrespective of clinical stage or CD4cell count, reduced infant mortalityby 76% and HIV progression by75%.5 Although EID has focusedon testing exposed infants betweenthe ages of 4 and 6 weeks, morerecently, birth testing has beenconsidered feasible.6 Birth testingusing polymerase chain reaction(PCR) can detect in utero infectionand allows for timeous initiationof treatment,7 which is vitallyimportant for these infants whohave rapid disease progressionand high mortality rates comparedwith infants who are infectedintrapartum.8

In 2013 and 2015, the SouthAfrican Department of Healthadopted updated guidelines forinfant HIV testing to identify allinfants infected with HIV as earlyas possible.3 The guidelines includedHIV PCR testing at 6 weeks of ageand testing of symptomatic orhigh-risk infants exposed to HIVany time after birth.9 Additionaltesting guidelines were added inJanuary 2015 to include PCR testingat birth for high-risk infants, then inJune 2015, birth HIV PCR testing of all

infants exposed to HIV togetherwith 10-week PCR testing wasrecommended.9,10

In this study, we tested the feasibilityof using models to identify high-risk newborns for intrauterineHIV infection for targeted PCRtesting. Parameters were extractedfrom individual or combinedclinical and laboratory data. Wecompare the number of newbornHIV infections identified usinga universal and a targeted HIVtesting approach at Kalafong

Provincial Tertiary Hospital (KPTH)in Pretoria, South Africa.

METHODS

A cross-sectional sample of newbornsexposed to HIV was recruited to thestudy. Data abstraction and HIVtesting occurred within 72 hoursof birth.

Data Collection Procedures

Trained research staff recruitedpatients. Researchers identifiedmothers with known HIV-positive or

FIGURE 1Consort diagram showing the selection of enrolled patients with HIV PCR test results. ART, anti-retroviral therapy.

2 DU PLESSIS et al by guest on June 22, 2019www.aappublications.org/newsDownloaded from

-negative status and mothers ofunknown HIV status daily fromMonday through Friday duringworking hours in the obstetric unitpre- and postdelivery. Mothers whowere HIV-negative were sent for HIVcounseling and testing if their last testwas done before 32 weeks’ gestation,in accordance with national protocols.

Researchers interviewed mothers andrecorded data onto a case reportform. Researchers reviewed medicalrecords for maternal HIV results,infant birth weight, duration of labor,and PMTCT interventions. Thefollowing infant parameters weredocumented: birth history;anthropometric data: weight, length,head circumference, and mid-upperarm circumference; and the presenceor absence of symptoms (seeSupplemental Information fordefinitions) that included failing tothrive (includes low birth weight[LBW]), birth weight #2.5 kg,congenital pneumonia,hepatosplenomegaly, oral candidiasis,significant lymphadenopathy, and anyopportunistic infections. Maternalcharacteristics documented includedantenatal care (ANC), labor history,PMTCT interventions and the timingthereof, the last documentedmaternal HIV viral load (VL), inaccordance with the current SouthAfrican PMTCT guideline,antiretroviral drug history, andcompliance (Supplemental Tables5–8). Poor compliance was defined as,95% dose compliance by using self-reported data or visual analog scale.

Research nurses or qualified medicaldoctors working at KPTH drew bloodfrom infants exposed to HIV. Bloodwas tested for HIV infection by usingtotal nucleic acid PCR.

The laboratory reported all HIVPCR–positive results to studyinvestigators on a daily basis.Newborns with positive orindeterminate HIV PCR results werereferred to HIV clinicians forspecialized care and managed by the

immunology (HIV) clinic inaccordance with national guidelines.Mothers with high VLs or othermedical problems were referred tothe relevant adult services.

Ethical and Legal Considerations

Management at KPTH grantedpermission to conduct the study, andthe University of Pretoria Faculty ofHealth Sciences Research EthicsCommittee gave ethical clearance(protocol 285_2014). Each mothergave written informed consent onenrollment in the study.

Case report forms were entereddirectly into the Research ElectronicData Capture (REDCap) (VanderbiltUniversity, Nashville, TN) systemhosted at the South Africa MedicalResearch Council.

Data Analysis

Data were analyzed by using thestatistical software SAS version 9.4TS1M5 (SAS Institute, Inc, Cary, NC).Continuous data were expressed asmeans and SDs or as medians andinterquartile ranges (IQRs) forskewed distributions. Discrete orcategorical data were summarized byusing frequencies and percentages.Normally distributed data werecompared by using independentt tests, otherwise nonparametricalternatives were used.

The weight-for-age z scores werecalculated by using 2000 Centers forDisease Control and Preventiongrowth charts adjusted for gestationalage for preterm infants and WorldHealth Organization 2006 growthcharts from the Multicentre GrowthReference Study for term infants. Weused univariate and multivariatelogistic regression models to identifyassociations between HIV infectionand maternal and infant parameters.The following predictor variableswere considered: preterm gestationalage (yes = 1, no = 0), LBW ,2.5 kg(yes = 1, no = 0), maternal VL value(lower than detectable level [LDL] =1, ,1000 = 2, $1000 = 3), maternal

VL value (LDL = 1, ,1000 = 2,1000–10 000 = 3, .10 000 = 4),maternal HIV seroconversion after 32weeks’ gestation (yes = 1, no = 0),maternal cART duration at birth,4 weeks (yes = 1, no = 0), small forgestational age (SGA) (yes = 1,no = 0), maternal tuberculosis (yes =1, no = 0), symptomatic (yes = 1,no = 0), and maternal CD4 cell countvalue (,200 cells per mm3 = 1,200–500 cells per mm3 = 2, .500cells per mm3 = 3). Multivariateregression was initially done ona saturated model (full model) andthereafter reduced to 2-, 3-, and 4-risk models. Model significance wasmeasured with P values of .05and .25.

After fitting logistic regressionmodels, the variables with the highestpredictive value were selected byusing Schwarz information criterionor Bayesian information criterion(BIC). The performance of theselected models was evaluated byusing the C-index or the area underthe receiver operating curve. Wedetermined the derived risk scores,sensitivity, and specificity as well asfalse-positive and false-negative ratesfor various cutoff values. Wedeveloped probability models withthe end point of a positive birth HIVPCR using logistic regression ofunivariate and multivariatecharacteristics and risk factors. Formodeling purposes, only positive andnegative PCR test results wereincluded.

RESULTS

Study Population

Between August 2014 and December2016, 15 175 live infants wereborn at KPTH, 3356 (22.12%) ofthese to mothers infected with HIV(Fig 1). Informed consent wasobtained from 1759 of 1911(92.05%) eligible patients. Patientswith birth HIV PCR test resultswere included.

PEDIATRICS Volume 143, number 6, June 2019 3 by guest on June 22, 2019www.aappublications.org/newsDownloaded from

Maternal Characteristics

Almost half of mothers who wereHIV positive were in their thirties(854 of 1759; 48.55%) and mosthad a secondary education (1557of 1759; 88.52%). Although mostwomen were not legally married(1370 of 1706; 80.30%), almosthalf received monetary support

from a partner (809 of 1726;46.87%) and just over a third wereemployed (599; 34.70%). Almost allmothers reported having access toa cell phone (1625 of 1720; 94.48%)and household assets such asa television (1415; 82.28%) andrefrigerator (1252; 72.79%)(Table 1).

Most mothers knew their HIV statusbefore delivery (98.77%) and wereon cART (1626 of 1704; 95.29%),namely fixed dose combination (FDC)treatment (1473; 90.09%), with 105(7.09%) having received FDC for,1 month before delivery. Close toa quarter of mothers who were HIV-positive (376 of 1687; 22.25%)

TABLE 1 Descriptive Statistics of Maternal, General, and HIV-Related Variables of the Cohort and of the Mother-Infant Pairs With Positive-Result HIV PCRTests, KPTH, August 2014 to December 2016

Variables Modalities Birth Cohorta n (%) PCR-Positive n of N (%) Pb (Positive Versus Negative)

General maternal informationMaternal age at delivery, y

(N = 1729)#20 81 (4.68) 5 of 81 (6.17) .0202

21–25 295 (17.06) 5 of 295 (1.69)26–29 401 (23.19) 11 of 401 (2.74)30–39 854 (49.39) 13 of 854 (1.52)$40 98 (5.67) 0 of 98 (00.00)

Level of education (N = 1638) No education 20 (1.22) 1 of 20 (5.00) .4087a

Primary 121 (7.39) 3 of 121 (2.48)Secondary 1495 (91.27) 30 of 1495 (2.01)Tertiary 2 (0.12) 0 of 2 (0.00)

Antenatal visits (N = 1566) No ANC visits 90 (5.75) 7 of 90 (7.78) .00051–2 visits 300 (19.16) 11 of 300 (3.67)

3–5 867 (55.36) 11 of 867 (1.27).5 341 (21.78) 3 of 341 (0.88)

Maternal HIV diagnosisMaternal HIV status known at

birth (N = 1630)Yes 1610 (98.77) 30 of 1610 (1.86) .0068a

No 20 (1.22) 3 of 20 (15.00)Seroconversion after 32 weeks’

gestation (N = 1618)Yes 360 (22.25) 10 of 360 (2.78) .2047a

No 1258 (77.75) 22 of 1258 (1.75)Maternal HIV treatment andcomplianceOn cART at birth (N = 1635) Yes 1558 (95.29)a 27 of 1558 (1.73) .0036a

No 77 (4.71) 6 of 77 (7.79)a1473 (90.09) of cART were

FDCcART duration ,4 wk at birth

(N = 1481)Yes 105 (7.09) 7 of 105 (6.67) .0020a

No 1377 (92.91) 20 of 1377 (1.45)Self-reported compliance to cART

(%)0–9 68 (5.27) 3 of 68 (4.41) .0312a

(N = 1290) 10–19 7 (0.54) 2 of 7 (28.57)Median = 96% 20–89 172 (13.33) 0 of 172 (0.00)IQR 1%–98% 90–100 1043 (80.85) 16 of 1043 (1.53)

Maternal VL and CD4 parametersMaternal VL #3 mo of delivery

(absolute value, copies permL) (N = 935)

LDL 562 (60.11) 1 of 562 (0.18) ,.0001

#1000 165 (17.65) 7 of 165 (4.24)$1000 208 (22.24) 27 of 208 (12.98)

Maternal CD4 count within 6 moof delivery (cells per mm3)(N = 1196)

,200 214 (17.89) 12 of 214 (5.61) .0089

200–500 588 (49.16) 12 of 588 (2.04).500 394 (32.94) 7 of 394 (1.78)

a Birth cohort includes all enrolled patients. Fisher or x2 calculations.b P values were calculated between the PCR-positive and PCR-negative study groups of the cohort.

4 DU PLESSIS et al by guest on June 22, 2019www.aappublications.org/newsDownloaded from

seroconverted toward the end ofpregnancy (.32 weeks’ gestation).Using laboratory data, we could traceVLs within 3 months of delivery for935 (53.16%) mothers. Similarly, wecould trace CD4 cell counts for 1196(67.99%) mothers for the preceding 6months. We recorded 562 (60.11%)mothers with undetectable VL (LDL),indicating virological control.Approximately 1 in 5 mothers (208;22.24%) had a VL level of .1000copies per mL documented within3 months before delivery. Weobserved mild immunologicsuppression in 588 (49.16%) and 394(32.94%) mothers who had CD4counts between 200 and 500 cells permm3 and .500 cells per mm3,respectively. Maternal VL andmaternal CD4 cell count values werenegatively correlated 20.103(P = .0053).

Few mothers had tuberculosis(21 of 1730; 1.21%), most of whom(13; 65%) were being treatedfor tuberculosis. Only 5.75% (97of 1688) of mothers had no ANCvisits, the median visits being 4(IQR 3–5 visits). Infants with positiveHIV PCR results were associatedwith mothers who had no ANCvisits (P = .0005) (Table 1).Infants with positive HIV PCRresults were associated with motherswho had mean VLs of 53 424(SD = 116854) copies per mL,whereas infants with negative HIVPCR results were associated withmothers who had mean VLs of 11 361

copies per mL (SD = 66185)(P = .0511).

Infant Characteristics

Of the infants enrolled in the study,53.16% (935) were boys. More thana quarter of enrolled infants (432 of1655; 26.10%) were born at ,38 (37completed) weeks, with a mediangestation of 35 weeks (IQR 32–36weeks) (Table 2).

Infants had a mean birth weight of2.83 kg. LBW (,2.5 kg) wasdocumented in 398 of 1598 (24.55%)of the infants who were HIVPCR–negative and 13 of 32 (40.63%)in the positive-result group(P = .0329 [x2]). The median z scorefor weight was 20.33 (IQR 21.06to 0.38).

Fewer than 15% (14.33%) of enrolledinfants displayed clinical symptomsat birth. No newborns hadgeneralized lymphadenopathy orextensive oral candidiasis. Evidenceof growth restriction or SGA wasdocumented in 204 of 1689 (12.08%)of the enrolled infants, of whom 6were PCR-positive. Pneumonia andanemia were documented in 1.7%and 0.3% of infants, respectively(Table 2). Newborns that weresymptomatic more frequently testedHIV-positive (P = .0042).

HIV PCR Results

Of the 1691 infants with birth PCRresults, 31 (1.8%) received HIV-positive results, 1646 (97.3%)received HIV-negative results, and 14

(0.83%) received indeterminateresults; 3.87% samples were rejected(68 of 1759). According to thenational HIV testing protocol, theindeterminate HIV PCR results wererepeated in 12 of the 14 patients; 2infants could not be traced. Half of therepeated HIV PCR tests (6 of 12)produced positive test results andhalf produced negative test results,increasing the positivity rate to 2.2%(37 of 1703).

The mean turnaround time for thePCR results was 68.27 hours (IQR46.05–93.88 hours).

HIV Risk Score Models

Univariate (Unweighted) Model

We used univariate regressionmodels to identify associationsbetween HIV PCR–positive outcomeand 10 infant and maternalcharacteristics.

Newborns with positive PCR resultswere significantly associated withmaternal VL levels ,1000 (odds ratio[OR] = 26.53; 95% confidenceinterval [CI] 1.353–520.310; P = .002)and $1000 (OR = 123.67; 95% CI7.385–.999.99); maternal cARTduration ,4 weeks (OR = 0.146, 95%CI 0.057–0.373; P , .0001); andsymptomatic newborns (P = .237,95% CI 0.100–0.561; P = .0011).

Multivariate Models

The saturated multivariate regressionmodel only identified 2 significantrisk factors (P , .05), namely

TABLE 2 Descriptive Statistics of Infant Characteristics of the Cohort and Infants Who Are PCR-Positive Born to Mothers Who Are HIV-Positive, KTPH, August2014 to December 2016

Variables Modalities Birth Cohort n (%) PCR-Positive n of N (%) P

Gestational age at birth, wk (N = 1588) ,34 105 (6.61) 3 of 105 (2.86) .0015 (Fisher)34–37 209 (13.16) 10 of 209 (4.78)$38 1526 (96.10) 19 of 1526 (1.25)

Any birth symptom and/or sign(N = 1689)

IUGR and/or SGA 204 (12.08) 6 of 204 (2.94) .0042

Pneumonia 28 (1.66) 1 of 28 (3.57)Anemia 5 (0.30) 0 (0.00)

Thrombocytopaenia 2 (0.12) 1 of 2 (50.00)Hepatomegaly 2 (0.12) 0 (0.00)Splenomegaly 1 (0.06) 1 of 1 (100.00)

IUGR, growth restriction.

PEDIATRICS Volume 143, number 6, June 2019 5 by guest on June 22, 2019www.aappublications.org/newsDownloaded from

maternal VL (P = .0002) andsymptomatic newborns (P = .02). Weretained variables with a P , .25 inadditional models to increasepredictive value; therefore, cART,4 weeks (P = .17) and SGA (P = .25)were added to risk models. These 4parameters were combined in 2-, 3-,and 4-risk regression models. Table 3illustrates risk models, starting withthe individual unweighted model,

followed by two 2-risk models(models 1 and 2), a 3-risk (model 3),and a 4-risk (model 4) model, andfinally the full model that used 9parameters in 1 weighted model.Maternal VL and infant symptomswere the first risks in the 2-riskmodel, whereas the second 2-modelscore modeled 2 maternalparameters: maternal VL andduration of maternal cART. The 3-risk

model incorporated maternal VL,cART duration, and infantsymptomatology; the 4-risk modeladded infant size for gestational ageto the mentioned 3 risks.

Probability Testing of Each Risk Model

Mothers with a VL $1000 copies permL who deliver a symptomaticnewborn have a 38% chance ofa positive PCR HIV test result at birth

TABLE 3 Associations Between Maternal and Infant Characteristics and Weighted Newborn HIV Acquisition in Infants Exposed to HIV in the Very EarlyInfant Diagnosis of HIV Study

Characteristics Unadjusted OR Adjusted OR Adjusted OR Adjusted OR Adjusted OR Adjusted OR

(95% CI) Model 1 Model 2 Model 3 Model 4 Full Model

P P P P P P

Preterm gestational age 0.55 (0.23–1.30) — — — — 0.51 (0.13–1.98).17 — — — — .33

LBW 0.42 (0.18–1.01) — — — — 2.11 (0.40–11.06).05 — — — — .38

Maternal VL (1) (LDL/,1000/$1000)Comparing values,1000 vs LDL 26.53 (1.35–520.31) 24.48

(1.27–472.49)30.92

(1.62–590.21)28.11

(1.50–527.33)27.84

(1.55–501.43)25.05

(1.74–359.98)$1000 vs LDL 123.67 (7.39 to

.999.99)103.17 (6.22 to

.999.99)141.50 (8.66 to

.999.99)117.69 (7.28 to

.999.99)113.82 (7.32 to

.999.99)100.45 (7.98 to

.999.99).0002 .0005 .0002 .0004 .0004 .0002

Maternal HIV seroconversion .32weeks’ gestation

0.71 (0.28–1.79) — — — — 0.81 (0.28–2.28)

.46 — — — — .68Maternal cART duration ,4 wk 0.15 (0.06–0.37) 0.31 (0.12–0.83) — 0.45 (0.16–1.29) 0.42 (0.14–1.24) 0.46 (0.15–1.39)

,.0001 .02 — .14 .12 .17SGA 0.48 (0.19–1.23) — — — 0.66 (0.22–1.97) 0.51 (0.16–1.60)

.12 — — — .46 .25Maternal TB 0.94 (0.05–18.32) — — — — 1.20 (0.04–39.56)

.97 — — — — .92Symptomatic 0.24 (0.10–0.56) — 0.18 (0.07–0.46) 0.21 (0.08–0.57) 0.24 (0.09–0.66) 0.22 (0.06–0.82)

.001 — .0004 .002 .006 .02Maternal CD4 cell count (,200/200–500/.500 cells per mm3)

— — — — — —

Comparing values,200 vs .500 0.35 (0.12–0.99) — — — — 0.83 (0.26–2.62)200–500 vs .500 0.31 (0.12–0.83) — — — — 0.71 (0.25–2.01)

.06 — — — — .81Maternal VL (2) (LDL/,1000/1000–10 000/.10 000)

— — — — — —

Comparing values,1000 vs LDL 26.528

(1.353–520.279)— — — — —

1000–10 000 vs LDL 76.990 (4.177 to.999.999)

— — — — —

.10 000 vs LDL 166.890 (9.799 to.999.999)

— — — — —

.0002 — — — — —

A total of 1600 observations were read, 33 of these from patients who are PCR-positive, and 634 observations were used. AIC/BIC/R2 for model 1: 135.08/152.89/0.08; model 2: 128.56/146.37/0.09; model 3:126.11/148.37/0.09; model 4: 125.16/151.87/0.10; final model: 123.86/177.29/0.10. AIC is an estimator of the relative quality of statistical models for a given set of data.Given a collection of models for the data, AIC estimates the quality of each model relative to each of the other models. BIC (SC), BIC/Schwarz criterion (also SBC, SBIC) is a criterion formodel selection among a finite set of models; the model with the lowest BIC is preferred (Cox-Snell measures). AIC, the Akaike information criterion; TB, tuberculosis; R2, variation in yexplained by the model; —, not applicable.

6 DU PLESSIS et al by guest on June 22, 2019www.aappublications.org/newsDownloaded from

(model 1–2 risks, Fig 2A). Another 2-risk model, containing only maternalcharacteristics, maternal VL and cARTduration, revealed a probability of28% for a newborn PCR-positive testresult if maternal VL is $1000 copiesper mL and the mother has notreceived cART treatment of.4 weeks(model 2–2 risks, Fig 2B). In model 3,newborns have a probability of 0.49of a positive-result PCR test if thenewborn is symptomatic, the motherhad received cART for ,4 weeksbefore delivery, and if the mother hada VL of $1000 copies per mL(Fig 2C). The 4-risk model includedmaternal characteristics VL and cARTduration and infant characteristicssymptomatic newborn and SGA; inthis model, newborns have thehighest probability of a positive HIVPCR result if maternal VL is $1000copies per mL, cART duration is

,4 weeks before delivery, and aninfant is symptomatic and SGA(model 4, Fig 2D).

Each model can be used at a riskscore probability cutoff to determinetargeted PCR testing algorithms.When using both the 3- and 4-riskmodel scores and consideringa probability of 0.02 and 0.04 as anindication for targeted birth testing,the sensitivity is 80% and 76%,respectively (Table 4).

DISCUSSION

In this study, we identifiedassociations between maternal andinfant characteristics and positive-result HIV PCR tests for newborns.We combined significant variables tobuild predictive models for earlyinfant HIV risk scores to detect HIV

infection at birth. According to ourmodels, newborns had the highest(0.57) probability for HIV PCRpositivity if newborns weresymptomatic and SGA and were bornto a mother who received cART for,4 weeks and had a VL $1000.Newborns had a 0.28 probability ofa positive-result HIV PCR test if onlymaternal VL $1000 and cART,4 weeks were included in themodel. Our study population had highHIV prevalence among mothers(22.12%), and overall PCR positivityin newborns was 2.2%, which ishigher than the national averageof 1.1%.3

Mothers who gave birth to newbornswith in utero–acquired HIV morefrequently had either no (P = .0023)or fewer than 3 ANC visits (P = .02).At the time of the study, basic ANC

FIGURE 2Probability of a newborn infected with HIV by risk factors using multivariable analysis as per models as illustrated in A–D. A, Model 1: 2 risks. B, Model 2:2 risks. C, Model 3: 3 risks (symptomatic infant). D, Model 4: 4 risks (SGA and symptomatic infant). art_less_4w_del, cART ,4 weeks’ duration;Mat_vl_grp, maternal VL group (1 = LDL, 2 = ,1000 copies per mL, 3 = equal and .1000 copies per mL).

TABLE 4 Sensitivity and Specificity Values of Both 3- and 4-Risk Models as Probability Levels 0.02 and 0.04

Probability Level True-Positive (Sensitivity) True-Negative(Specificity)

3- and 4-risk scores 0.02 20 of 25 (80) 494 of 773 (64)0.04 19 of 25 (76) 603 of 773 (78)

PEDIATRICS Volume 143, number 6, June 2019 7 by guest on June 22, 2019www.aappublications.org/newsDownloaded from

guidelines in South Africarecommended an early ANCvisit (,12 weeks’ gestation) followedby 4 follow-up ANC visits for low-riskpregnancies. In mid 2016, 8 follow-upvisits were recommended, andmothers who test HIV-positive duringpregnancy should initiate cARTa week later.11 Mothers whoare infected with HIV shouldreceive continuous educationabout pregnancy-related issues.ANC should complement HIVmonitoring and treatment in thesemothers.

Almost all mothers knew their HIVstatus before delivery (99%) andwere on cART (95%). Encouragingly,these values are in line with 2 of thecurrent United Nations Programmeon HIV/AIDS 90-90-90 goals; namely,90% of all patients infected with HIVshould be aware of their HIV status,and 90% of patients infected withHIV should receive cART.12 Motherswho had been on cART for shorterthan 4 weeks were significantlyassociated with newborns who hada positive-result PCR HIV test at birth.Despite most mothers knowing theirstatus and receiving cART, only 60%had achieved viral suppression, farless than the target of 90% of HIVpatients on cART who should bevirally suppressed to reach the 90-90-90 goals.

In utero HIV infection in newbornswas significantly associated witha detectable maternal VL. Both VLlevels ,1000 and $1000 copies permL proved significant risks for inutero HIV acquisition (P = .0002 andP , .0001, respectively) in bothunivariate and multivariate analyses.Myer et al13 emphasized theimportance of diligent maternal VLmonitoring and management duringpregnancy and breastfeeding. VLtesting 4 weeks before or at deliverycan provide valuable information toguide targeted interventions frombirth, including birth PCR testing.13,14

Health care facilities providing ANCservices should focus on the

importance of providing cART formothers at least 4 weeks beforedelivery. Mothers who are HIV-positive should be enrolled in PMTCTprograms, which should considerrepeat HIV testing, treatment, and VLmonitoring at 34 to 36 weeks’gestation. In the future, the value ofmaternal VL point-of-care testingshould be assessed.

In our study, newborns infected withHIV were more likely to be preterm(,38 weeks’ gestation), LBW, andSGA, although none of theseparameters were significantlyassociated with a positive-result HIVPCR test in the univariate regressionmodels. We included SGA values inthe 4-model multivariate analysis,which added predictive value totargeted testing models. Werecommend that all health carefacilities with maternity services haveinfant gestational age tables andinfant scales to determine weight-for-age measurements for newbornsexposed to HIV.

The newborn HIV PCR positivityrate was 1.83% (31 of 1691)with ,5% of samples rejected (68,3.9%). After repeat testing of theindeterminate results, an additional6 patients tested HIV-positive,increasing the positivity rate to2.2% (37 of 1703). With a maternalHIV positivity percentage of 22% and1.8% mother-to-child transmissionof HIV (MTCT), 396 per 100 000live births were HIV infected inthis study. Goga et al15 looked atearly (4–8 weeks) MTCT risk inSouth Africa from 2011–2012 to2012–2013 in relation to theSouth African PMTCT strategy,which is in line with World HealthOrganization Option A. In2011–2012, MTCT was 2.7% and2.6% in 2012–2013, varying betweenprovinces (1.9%–5.4%). Mothers whostarted cART during or before thefirst trimester of pregnancy hadlow risk of early MTCT (1.2%).2 Werecorded a similar rate of MTCT,indicating that most MTCT events are

therefore in utero and intrapartum.The risk has not changed significantlyin the past 5 to 6 years. The currentSouth African MTCT case rate is notclose to the elimination of MTCTtarget of #50 per 100 000 livebirths.16

CONCLUSIONS

In this study, we present models fortargeted birth PCR testing, which maybe useful in resource-constrainedsettings. Our findings indicate thatmaternal VL testing is vital fora targeted birth PCR approach. Our 3-and 4-risk models achieveda sensitivity of close to 80%,indicating that 1 in 5 newbornsinfected with HIV will not be offeredtargeted birth testing. Globally, EID,early treatment, and elimination ofMTCT are prioritized; thus, wesupport universal birth testing withinthe South African PMTCT program.

ACKNOWLEDGMENT

Language editing was done by CherylTosh, University of Pretoria.

ABBREVIATIONS

ANC: antenatal careBIC: Bayesian information

criterioncART: combined 3-drug

antiretroviral therapyCI: confidence intervalEID: early infant diagnosisFDC: fixed dose combinationIQR: interquartile rangeKPTH: Kalafong Provincial Tertiary

HospitalLBW: low birth weightLDL: lower than detectable levelMTCT: mother-to-child

transmission of HIVOR: odds ratioPCR: polymerase chain reactionPMTCT: prevention of mother-to-

child transmission of HIVSGA: small for gestational ageVL: HIV viral load

8 DU PLESSIS et al by guest on June 22, 2019www.aappublications.org/newsDownloaded from

final manuscript; Prof Goga conceptualized and designed the methodology, assisted with developing the software for data collection, collected data, drafted the

original manuscript, and reviewed and edited the final manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all

aspects of the work.

DOI: https://doi.org/10.1542/peds.2018-3834

Accepted for publication Feb 8, 2019

Address correspondence to Nicolette M. Du Plessis, PhD, Department of Paediatrics, University of Pretoria, Pretoria 0001, South Africa. E-mail: nicolette.duplessis@

up.ac.za

PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).

Copyright © 2019 by the American Academy of Pediatrics

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

FUNDING: The South African Medical Research Council employed 2 dedicated research nurses for the study.

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2019-0913.

REFERENCES

1. Shetty AK, Maldonado Y. Preventingmother-to-child transmission of humanimmunodeficiency virus type 1 inresource-poor countries. Pediatr InfectDis J. 2003;22(6):553–555

2. Goga AE, Dinh TH, Jackson DJ, et al;South Africa PMTCT Evaluation(SAPMCTE) Team. Population-leveleffectiveness of PMTCT Option A onearly mother-to-child (MTCT)transmission of HIV in South Africa:implications for eliminating MTCT.J Glob Health. 2016;6(2):020405

3. Moyo F, Haeri Mazanderani A, Barron P,et al. Introduction of routine HIV birthtesting in the South African NationalConsolidated Guidelines. Pediatr InfectDis J. 2018;37(6):559–563

4. Biggar RJ, Mtimavalye L, Justesen A,et al. Does umbilical cord bloodpolymerase chain reaction positivityindicate in utero (pre-labor) HIVinfection? AIDS. 1997;11(11):1375–1382

5. Violari A, Cotton MF, Gibb DM, et al;CHER Study Team. Early antiretroviraltherapy and mortality among HIV-infected infants. N Engl J Med. 2008;359(21):2233–2244

6. Essajee S, Bhairavabhotla R, PenazzatoM, et al. Scale-up of early infant HIVdiagnosis and improving access topediatric HIV care in global plancountries: past and future perspectives.J Acquir Immune Defic Syndr. 2017;75(suppl 1):S51–S58

7. Lilian RR, Kalk E, Bhowan K, et al. Earlydiagnosis of in utero and intrapartumHIV infection in infants prior to 6 weeksof age. J Clin Microbiol. 2012;50(7):2373–2377

8. Zijenah LS, Moulton LH, Iliff P, et al;ZVITAMBO Study Group. Timing ofmother-to-child transmission of HIV-1and infant mortality in the first6 months of life in Harare, Zimbabwe.AIDS. 2004;18(2):273–280

9. Department of Health. The SouthAfrican antiretroviral treatmentguidelines 2013: PMTCT guidelines.2013. Available at: www.kznhealth.gov.za/medicine/2013_art_guidelines.pdf.Accessed October 20, 2018

10. National Department of Health. NationalConsolidated Guidelines for thePrevention of Mother-to-ChildTransmission of HIV (PMTCT) and theManagement of HIV in Children,Adolescents and Adults. Pretoria, SouthAfrica: National Department of Health;2015. Available at: www.kznhealth.gov.za/family/HIV-Guidelines-Jan2015.pdf.Accessed October 23, 2018

11. Department of Health. Guidelines formaternity care in South Africa. 4th ed.2015. Available at: https://www.medbox.org/guidelines-for-maternity-care-in-south-africa/preview?q=guidelines-for-maternity-care-in-south-africa.Accessed October 13, 2018

12. UNAIDS. 90-90-90: an ambitioustreatment target to help end the AIDS

epidemic. Geneva, Switzerland: UNAIDS.2014. Available at: http://www.unaids.org/sites/default/files/media_asset/90-90-90_en.pdf. Accessed August 20, 2018

13. Myer L, Essajee S, Broyles LN, et al.Pregnant and breastfeeding women:a priority population for HIV viral loadmonitoring. PLoS Med. 2017;14(8):e1002375

14. World Health Organization.Consolidated Guidelines on the Use ofAntiretroviral Drugs for Treating andPreventing HIV Infection. Geneva,Switzerland: WHO; 2016. Available at:www.who.int/hiv/pub/arv/arv-2016/en/.Accessed July 20, 2018

15. Goga AE, Dinh TH, Jackson DJ;SAPMTCTE Study Group. Evaluation ofthe Effectiveness of the NationalPrevention of Mother-To-ChildTransmission (PMTCT) ProgrammeMeasured at Six Weeks Postpartum inSouth Africa, 2010. Pretoria, SouthAfrica: South African Medical ResearchCouncil, National Department of Healthof South Africa and PEPFAR/US Centersfor Disease Control and Prevention;2012

16. World Health Organization. Globalguidance on criteria and processes forvalidation: elimination of mother-to-child transmission of HIV and syphilis.2017. Available at: https://www.who.int/reproductivehealth/publications/rtis/9789241505888/en/. Accessed October23, 2018

PEDIATRICS Volume 143, number 6, June 2019 9 by guest on June 22, 2019www.aappublications.org/newsDownloaded from

DOI: 10.1542/peds.2018-3834 originally published online May 17, 2019; 2019;143;Pediatrics 

Ameena E. GogaNicolette M. Du Plessis, Chris J.B. Muller, Theunis Avenant, Michael S. Pepper and

An Early Infant HIV Risk Score for Targeted HIV Testing at Birth

ServicesUpdated Information &

http://pediatrics.aappublications.org/content/143/6/e20183834including high resolution figures, can be found at:

Referenceshttp://pediatrics.aappublications.org/content/143/6/e20183834#BIBLThis article cites 9 articles, 1 of which you can access for free at:

Subspecialty Collections

http://www.aappublications.org/cgi/collection/hiv:aids_subHIV/AIDSbhttp://www.aappublications.org/cgi/collection/infectious_diseases_suInfectious Diseasefollowing collection(s): This article, along with others on similar topics, appears in the

Permissions & Licensing

http://www.aappublications.org/site/misc/Permissions.xhtmlin its entirety can be found online at: Information about reproducing this article in parts (figures, tables) or

Reprintshttp://www.aappublications.org/site/misc/reprints.xhtmlInformation about ordering reprints can be found online:

by guest on June 22, 2019www.aappublications.org/newsDownloaded from

DOI: 10.1542/peds.2018-3834 originally published online May 17, 2019; 2019;143;Pediatrics 

Ameena E. GogaNicolette M. Du Plessis, Chris J.B. Muller, Theunis Avenant, Michael S. Pepper and

An Early Infant HIV Risk Score for Targeted HIV Testing at Birth

http://pediatrics.aappublications.org/content/143/6/e20183834located on the World Wide Web at:

The online version of this article, along with updated information and services, is

http://pediatrics.aappublications.org/content/suppl/2019/05/16/peds.2018-3834.DCSupplementalData Supplement at:

ISSN: 1073-0397. 60007. Copyright © 2019 by the American Academy of Pediatrics. All rights reserved. Print the American Academy of Pediatrics, 141 Northwest Point Boulevard, Elk Grove Village, Illinois,has been published continuously since 1948. Pediatrics is owned, published, and trademarked by Pediatrics is the official journal of the American Academy of Pediatrics. A monthly publication, it

by guest on June 22, 2019www.aappublications.org/newsDownloaded from