Home-based HIV Testing for Children: A Useful Complement for Caregivers with More Children, Who are Male, and with an HIV Negative Partner

Abstract

Expanding index and family-based testing (HBT) is a priority for identifying children living with HIV. Our study characterizes predictors that drive testing location choice for children of parents living with HIV. Kenyan adults living with HIV were offered a choice of HBT or clinic-based testing (CBT) for any of their children (0–12 years) of unknown HIV status. Multi-level generalized linear models were used to identify correlates of choosing HBT or CBT for children and testing all versus some children within a family, including caregiver demographics, HIV history, social support, cost, and child demographics and HIV prevention history. Among 244 caregivers living with HIV and their children of unknown HIV status, most (72%) caregivers tested children using CBT. In multivariate analysis, female caregivers [aRR 0.52 (95% CI 0.34–0.80)] were less likely to choose HBT than male caregivers. Caregivers with more children requiring testing [aRR 1.23 (95% CI 1.05–1.44)] were more likely to choose HBT than those with fewer children requiring testing. In subgroup univariate analysis, female caregivers with a known HIV negative spouse were significantly more likely to choose HBT over CBT than those with a known HIV positive spouse [RR 2.57 (95% CI 1.28–5.14), p = 0.008], no association was found for male caregivers. Child demographics and clinical history was not associated with study outcomes. Caregiver-specific factors were more influential than child-specific factors in caregiver choice of pediatric HIV testing location. Home-based testing may be preferable to families with higher child care needs and may encourage pediatric HIV testing if offered as an alternative to clinic testing.

Introduction

Globally, there are an estimated of 1.7 million children living with HIV, and over 90% of these children reside in sub-Saharan Africa [1, 2]. Prevention of mother-to-child transmission of HIV (PMTCT) programs, which began scaling up in the early 2000s, have reduced the number of new infant HIV infections and included routine testing for early infant diagnosis (EID). However, low HIV testing coverage for children contributes to low coverage of antiretroviral therapy treatment (ART) for children, at 54% compared to 74% for adults globally in 2020 [3]. Expanding rights-based index, family, and household testing to diagnose and link children to treatment is a priority for next 5 years by UNAIDS [1].

While Kenyan national guidelines recommend outpatient provider-initiated testing (PITC) for all children when they present to a clinic, barriers and gaps in testing coverage persist [4]. Untreated children living with HIV have high morbidity and mortality [5-7], and delayed diagnosis and treatment until children are critically ill is associated with substantially higher mortality [8]. Thus, identifying children living with HIV early and linking them to treatment before they are symptomatic and hospitalized is critical to improve health outcomes.

Compared with community-based door-to-door testing or outreach testing, index case testing—identifying and testing children of parents or of siblings living with HIV—is more efficient in identifying children who are HIV-infected before they become symptomatic [4, 9, 10]. A study in Kenya found an HIV prevalence of 4.5% among children aged 0–14 years, who were tested at home or at clinic through index case testing, higher than the yield in both outpatient (1.2%) and inpatient (1.6%) PITC [10]. In a meta-analysis of the index case testing strategy, pooled HIV prevalence was 8.4%, but uptake of testing was sub-optimal at 52% [4]. Understanding the co-factors related to caregivers’ testing decisions may increase uptake of index case testing and accelerate the identification of undiagnosed children.

Home-based testing (HBT) has been offered to adults extensively and shows potential to reach often-missed populations, such as first-time testers and index case testing of family members [11-14]. Among adults, gender, age, relationship status, and HIV history have been associated with choosing or not choosing HBT, although the direction of these associations have been mixed in different studies [14-16]. Testing children for HIV is further complicated by issues of consent, autonomy, delivery of results, and caregiver disclosure issues [17]. In Kenya, a parent or legal guardian (referred to as caregivers in this study) can consent for the child(ren) in their care to receive an HIV test; older child and adolescent (ages 10–17) assent should also be obtained [18]. Little is known about cofactors of choosing HBT for index case pediatric testing. Thus, evidence is needed to inform targeted HIV testing program decisions for children.

We aimed to determine caregiver- and child-level cofactors associated with pediatric HIV testing at home versus at clinic, to determine caregiver-level factor associated with complete testing for all eligible children, and to determine child-level factors associated with which child(ren) gets tested.

Methods

Study Design

This study used data collected in the CATCH study in Kenya from 2013 to 2015. Study design, recruitment, and enrollment have been described in detail previously [19, 20]. Briefly, adults living with HIV (ALHIV) seeking HIV care were systematically screened to identify those with any children of unknown HIV status ages 0–12 years and were offered enrollment into the study. Adults who were biological parents or guardians of children were included; these included a range of family relationships including biological parents, grandparents, aunts/uncles, etc. All adults completed a self-report enrollment questionnaire, including caregiver demographics [including caregiver sex, age, number of untested and number of deceased children, relationship status (categorized as having no spouse, an unsupportive spouse, or a supportive spouse)], caregiver and spouse HIV history [including when caregivers started HIV care and ART, and spouse HIV status (categorized as known positive, known negative, or unknown status)], caregiver social support (including self-reported disclosure to spouse, concerns about inadvertent disclosure, needing spouse permission to test children, perceived stigma for child, concerns about child finding out about their positive status, and perception that the child was not sick), economic factors (including income, home ownership, rent, need to miss work, and health insurance), travel factors [including cost, distance (calculated using GPS coordinates of the nearest landmark to the home), and travel time to HIV care clinic], and child-level demographic and clinical factors (including sex, age, breastfeeding, PMTCT, and hospitalization history).

At the end of the enrollment visit, caregivers were given a choice of whether to test their children at home or to bring their children to the clinic for HIV testing, either on the same day or at a later date. Caregivers with accompanying children who tested their children in the clinic the same day that they were approached to join the CATCH study were excluded from this analysis.

Study Setting

Caregivers were recruited from 3 public hospitals and 4 health centers in Nairobi and Kisumu, Kenya. Nairobi is a high-density urban area with smaller family size and lower HIV prevalence than in Kisumu. Kisumu is a primarily urban environment, but recruitment clinics attracted patients from more outlying, rural regions as well. In 2018, the adult HIV prevalence in Nairobi and Kisumu was 6.1% and 16.3%, respectively [21]. In 2012, when CATCH started, a national survey in Kenya revealed that only 41% of the children living with HIV 18 months to 14 years were diagnosed with HIV [22].

Theoretical Model

We identified a range of hypothesized caregiver- and child-level factors associated with caregiver’s pediatric HIV testing preferences using an adapted Andersen’s model for health services utilization specifically adapted for pediatric HIV testing [23]. We considered the following caregiver-level factors: caregiver demographics, caregiver partnership status, caregiver and spouse’s HIV history, social support, distance to HIV care clinic, and economic factors; and the following child-level factors: child demographics, HIV exposure and prevention history.

Statistical Analysis

We assessed 3 different outcomes: (1) whether a child was tested via CBT or HBT (child-level outcome), (2) among families that had ≥ 1 untested children and tested at least one child, whether all or not all children were tested (family-level outcome), and (3) among families that did not test all children, which children were tested (child-level outcome).

We utilized multilevel generalized linear regression models (GLM) with a log link and binomial family, adjusted for sites, and accounting for clustering at the family level for analyses with child-level outcomes, and calculated the relative risks (RR) and 95% confidence interval (95% CI) for each correlates. We conducted multivariate analysis using the same modeling techniques, including only those cofactors that had a p-value < 0.1 in univariate analyses and excluded those that were collinear (in which the standard error changed by > 10% when included in the same model). All analyses were conducted using Stata 14 (StataCorp, College Statin, TX).

Ethical Considerations

This study was approved by the University of Washington Institutional Review Board and the Kenyatta National Hospital/ University of Nairobi Ethics and Research Committee. All adults provided written informed consent for enrollment.

Results

Among 493 caregivers enrolled, 492 agreed to test their eligible children, of whom 103 (20.9%) had their child with them at enrollment and completed testing in the clinic the same day and were excluded from additional analyses described here. Of the remaining 389 caregivers, 145 (37.3%) never tested their child(ren), 175 (45.0%) brought their child(ren) to the clinic for CBT, and 69 (17.7%) had their child(ren) tested using HBT. Among the 244 caregivers who tested children, 72% completed CBT.

Demographic Characteristics

A majority of the 244 caregivers included in this analysis were female (84%), the median age was 34, had completed primary education (median 9 years of education), were fairly recently diagnosed with HIV (median 2 years since diagnosis), and were currently receiving ART (75%). Among caregivers who had a spouse, 55% of their spouses were known to be living with HIV. More than half of the caregivers had more than one child of unknown HIV status (median 2 untested children) and 5% of the caregivers reported having children of known HIV-positive status. Male caregivers were significantly more likely than female caregivers to have more children [median (IQR) 3 (2–3) vs. 2 (1–3); p = 0.0022] and more untested children [median (IQR) 2 (2–3) vs. 2 (1–3); p = 0.0003]. Among 376 children who were tested, the majority (58%) were female, the median age was 7 years, the median duration breastfed was 12 months, and 5% had been hospitalized previously (Table 1).

Table 1.

Baseline characteristics of caregivers living with HIV and their children

	n (%) or median (IQR)
Caregiver demographics (N = 244)
Female	206 (84.4)
Age	34 (28–39)
Years of education	9 (8–12)
Marital status
Single (never married)	19 (7.8)
Divorced/separated	41 (16.8)
Widowed	27 (11.1)
Married monogamous	136 (55.7)
Married polygamous	18 (7.4)
Steady boyfriend	2 (0.8)
Othera	1 (0.4)
HIV history
Years since diagnosis	2 (0–5)
On ART	182 (75.2)
Has spouseb (N = 154)
Spouse known HIV positive	85 (54.5)
Spouse known HIV negative	22 (14.1)
Spouse HIV status unknown	47 (30.1)
Family characteristics
Families with known HIV positive living children	12 (4.9)
Families with deceased children	43 (17.6)
Total number of children per family ages 0–12 years oldc	2 (1–3)
Untested children per family ages 0–12 years oldc	2 (1–2)
City of residence
Nairobi	200 (82.0)
Kisumu	44 (18.0)
Child demographics (N = 376)
Female	217 (57.7)
Age (years)	7 (5–10)
Months breastfed	12 (8–24)
Has been hospitalized	12 (4.9)
Child PMTCT history
Mother tested negative or not tested	232 (61.7)
Mother HIV positive and no EID testing	24 (6.4)
Mother HIV positive and child EID tested	33 (8.8)
Unknown maternal and/child testing behavior or result	87 (23.1)
PMTCT history for children from female caregivers (N = 307)
Mother tested negative or not tested	212 (69.1)
Mother HIV positive and no EID testing	23 (7.5)
Mother HIV positive and child EID tested	30 (9.8)
Unknown maternal and/child testing behavior or result	42 (13.7)
PMTCT history for children from male caregivers (N = 69)
Mother tested negative or not tested	20 (29.0)
Mother HIV positive and no EID testing	1 (1.4)
Mother HIV positive and child EID tested	3 (4.3)
Unknown maternal and/child testing behavior or result	45 (65.2)

^aOther: One female caregiver self-reported she was widowed and was inherited by another man after the death of husband

^bAmong 156 with any kind of partner, 2 had steady boyfriends and were excluded from the category of having a spouse, which was restricted to the 154 married caregivers

^cFor caregivers included in this analysis (N = 244) who tested at least one of their untested children under 12, male caregivers (N = 38) had 3 (2–3) total children and 2 (2–3) untested children. Female caregivers (N = 206) had 2 (1–3) total children and 2 (1–3) untested children. This was statistically significantly different by t-test (total children p = 0.0022, untested children p = 0.0003)

PMTCT history records were more complete for children with female versus male caregivers (14 vs. 65% unknown information, respectively). Among female caregivers, 69% reported that the child’s mother tested HIV negative or was of unknown HIV status during pregnancy, 8% of the child’s mother was HIV positive but that the child did not receive any early infant diagnosis testing, and 10% of the child’s mother was HIV positive, and that the child received some, but not all, recommended early infant diagnosis testing (Table 1).

Caregiver- and Child-Level Cofactors of CBT vs. HBT

A total of 376 children were tested; 253 (67%) were tested at clinic and 123 (33%) were tested at home. Caregiver demographics were associated with child testing location. Female caregivers were less likely to choose HBT than male caregivers [RR 0.54 (95% CI 0.34–0.86)]. Caregivers with more children of unknown HIV status were more likely than those with fewer children to choose HBT [RR 1.24 (95% CI 1.06–1.46)]. Compared to caregivers with unsupportive spouses, caregivers with no spouses were less likely to choose HBT [RR 0.38 (95% CI 0.17–0.88)]. Caregiver’s economic factors, travel factors, and child-level factors were not significantly associated with testing location preference (Table 2).

Table 2.

Family- and child-level cofactors associated with completing home-based testing

Cofactors	Children tested by CBT (N = 253)		Children tested by HBT (N = 123)		RRg	95% CI	p-value	aRRg	95% CI	p-value
	N	n (%) or median (IQR)	N	n (%) or median (IQR)
Caregiver demographics
Female	253	220 (87.0)	123	87 (70.7)	0.54	0.34–0.86	0.010	0.52	0.34–0.80	0.003
Age (years)	253	33 (29–38)	123	34 (29–39)	1.01	0.98–1.04	0.453
Number of children of unknown HIV status aged 0–12 y/o	253	2 (1–3)	123	2 (2–3)	1.24	1.06–1.46	0.009	1.23	1.05–1.44	0.010
Has deceased children	253	47 (18.6)	123	25 (20.3)	1.14	0.68–1.92	0.626
Relationship status a	249		123
No spouse		57 (22.9)		8 (6.5)	0.38	0.17–0.88	0.023
Supportive spouse		132 (53.0)		88 (71.5)	1.28	0.80–2.06	0.307
Unsupportive spouse		60 (24.1)		27 (22.0)	Ref	Ref	Ref
Caregiver and spouse's history with HIV
Years since caregiver diagnosed	253	1 (0–5)	123	2 (0–4)	1.00	0.95–1.06	0.961
Caregiver on ART	253	182 (72.3)	123	97 (78.9)	1.35	0.82–2.22	0.233
Spouse HIV status b	160		102
Known HIV positive		94 (58.8)		56 (54.9)	Ref	Ref	Ref
Known HIV negative		14 (8.8)		23 (22.5)	1.62	0.92–2.86	0.094	1.99	1.20–3.30	0.008
Unknown		55 (34.4)		23 (22.5)	0.76	0.44–1.23	0.316	0.99	0.60–1.63	0.965
Caregiver's social support
Disclosure to spouse	157	140 (89.2)	102	98 (96.1)	2.15	0.750–6.177	0.154
Concerns for inadvertent disclosure of caregiver HIV status	253	109 (43.1)	122	58 (47.5)	1.01	0.66–1.55	0.957
Need spouse permission to test child	160	25 (15.6)	102	15 (14.7)	0.86	0.459–1.617	0.643
Perceived stigma at school or community for child	253	111 (43.9)	123	79 (64.2)	1.47	0.94–2.31	0.090	1.53	0.97–2.42	0.070
Concerns for child find out being positive	253	134 (53.0)	123	76 (61.8)	1.03	0.65–1.62	0.909
Child perceived not being sick	171	35 (20.5)	76	33 (43.4)	1.28	0.71–2.34	0.414
Economic factors
Has income	253	178 (70.4)	123	86 (69.9)	1.06	0.67–1.67	0.809
Monthly income (per 10 USD)e	178	$78 ($52–$130)	86	$104 ($65–$273)	1.01	0.99–1.02	0.211
Owns home	253	22 (8.3)	123	7 (5.7)	1.10	0.45–2.68	0.841
Monthly rent (per 10 USD)e	231	$32 ($19—$71)	116	$33 ($16—$52)	1.03	0.98–1.07	0.224
Time to travel to clinic (per 10 min)	253	35 (20–60)	121	45 (25–60)	0.99	0.95–1.03	0.638
Caregiver home distance to clinic (km)	105	8.1 (2.6–19.3)	59	12.7 (4.9–21.2)	0.99	0.97–1.01	0.535
Caregiver has cost to travel to clinic	253	158 (62.5)	123	94 (77.2)	1.35	0.84–2.17	0.214
Caregiver travel fare (per 1 USD) †	158	$1.10 ($0.65-$1.95)	94	$1.30 ($0.65-$1.95)	1.00	0.94–1.06	0.958
Children has travel cost to clinic	253	167 (66.0)	123	88 (71.5)	1.18	0.75–1.84	0.472
Travel cost if bring all children to clinic (per 1 USD)e	167	$2.08 ($1.17—$3.12)	88	$2.60 ($1.30–$5.07)	1.08	0.96–1.20	0.191
Miss work to test child	253	100 (39.5)	123	59 (48.0)	1.24	0.82–1.88	0.311
Has NHIF cardc	253	83 (32.8)	123	54 (43.9)	1.25	0.82–1.91	0.301
Child demographics and clinical history
Female	253	139 (54.9)	123	78 (63.4)	1.30	0.88–1.91	0.192
Age (years)	253	7 (5–10)	123	7 (5–9)	0.99	0.93–1.05	0.720
Months breastfed	251	14 (8–24)	117	12 (8–20)	0.99	0.97–1.01	0.390
Child PMTCT d	220		87
Mother tested negative or not tested		150 (67.7)		62 (71.3)	Ref	Ref	Ref
Mother HIV positive and no EID testing		19 (8.6)		4 (4.6)	0.63	0.22–1.82	0.397
Mother HIV positive and child EID tested		18 (8.2)		12 (13.8)	0.99	0.50–1.97	0.983
Has been hospitalizedf	175	12 (6.9)	68	0 (0)	–	–	0.999

^aSpouse supportiveness assessed using the question, “Have you experienced positive support regarding your HIV status from your partner?”; This variable excluded from multivariate analyses due to collinearity

^bIn univariate analyses stratified by caregiver sex, there was evidence of effect modification. Female caregivers with a known HIV negative spouse were significantly more likely to choose HB over CBT than those with a known HIV positive spouse [RR 2.57 (95% CI 1.28–5.14), p = 0.008]. However, the opposite direction association was observed, although not statistically significantly, among male caregivers [RR 0.54 (95% CI 0.13–2.29), p = 0.407]. Neither among female nor male caregivers was there a substantial or significant association between having a spouse of unknown HIV status and choice of HBT over CBT [female: RR 0.96 (95% CI 0.49–1.89), p = 0.897; male: RR 1.26 (95% CI 0.48–3.32), p = 0.643]

^cNHIF card: National Health Insurance Fund card in Kenya

^dUnknown testing behaviors and status (n = 42; 13.7%) from female caregivers were not included for univariate analysis. Responses from male caregivers were excluded due to missing data

^eData collected in 2014, all monetary values are converted and inflated from 2014 Kenya shillings to 2020 US dollars values

^fVariable collected after a certain date; missing data is likely at random

^gRR adjusted for sites and clustered on family

In multivariate analysis including correlates that had a p-value < 0.1 in univariate analyses that were not collinear (caregiver relationship status was excluded due to collinearity), female caregivers [aRR 0.52 (95% CI 0.34–0.80)] remained significantly less likely to choose HBT; while caregivers with more untested children [aRR 1.23 (95% CI 1.05–1.44)], and those with HIV negative spouses [aRR 1.99 (95% CI 1.20–3.30)] compared those with HIV positive spouses, were more likely to choose HBT (Table 2).

Furthermore, we found caregiver sex has an effect modification to testing choice in subgroup analysis where we assessed the relationship between spouse HIV status and choice of HBT versus CBT stratified by male and female caregivers. Female caregivers with a known HIV negative spouse were significantly more likely to choose HBT over CBT than those with a known HIV positive spouse [RR 2.57 (95% CI 1.28–5.14), p = 0.008]. However, the opposite direction association was observed, although not statistically significantly, among male caregivers [RR 0.54 (95% CI 0.13–2.29), p = 0.407]. Neither among female nor male caregivers were there a substantial or significant association between having a spouse of unknown HIV status and choice of HBT over CBT [female: RR 0.96 (95% CI 0.49–1.89), p = 0.897; male: RR 1.26 (95% CI 0.48–3.32), p = 0.643].

Caregiver-Level Factors for Testing All Versus not all Children of Unknown-Status

A total of 323 children were from 125 families who had at least one child tested in CATCH. Among these 125 families, 50 (40%) did not test all of their children of unknown status and 75 (60%) completed testing for all children. Caregivers who had more children of unknown status were less likely to test all of their children than caregivers with fewer children [RR 0.55 (95% CI 0.37–0.81)] for each additional child who needed testing). Caregiver’s HIV history, social support, stigma, and economic factors were not associated with caregivers’ decisions to complete testing for all children (Table 3).

Table 3.

Caregiver-level cofactors for testing all or not all eligible children within the study

Cofactors	Family where not all chil dren were tested (N = 151)		Family where all children were tested (N = 172)		RRc	95% CI	p-value
	N	n (%) or median (IQR)	N	n (%) or median (IQR)
Caregiver demographics
Female	50	38 (76.0)	75	58 (77.3)	1.03	0.60–1.77	0.913
Age (years)	50	31.5 (27–39)	75	32 (29–39)	1.01	0.98–1.04	0.523
Number of children of unknown status aged 0–12 y/o	50	3 (2–4)	75	2 (2–3)	0.55	0.37–0.81	0.003
Has deceased children	50	9 (18.0)	75	16 (21.3)	1.09	0.62–1.90	0.768
Relationship status	49		75
No spouse		8 (16.3)		7 (9.3)	0.93	0.38–2.29	0.872
Supportive spouse		26 (53.1)		53 (70.7)	1.34	0.76–2.39	0.314
Unsupportive spouse		15 (30.6)		15 (20.0)	Ref	Ref	Ref
Years since caregiver diagnosed	50	1.5 (0–5)	75	0 (0–4)	0.99	0.93–1.06	0.767
Caregiver on ART	50	43 (86.0)	75	53 (70.7)	0.73	0.44–1.20	0.210
Spouse HIV status	36		59
Known HIV positive		19 (52.8)		29 (49.2)	Ref	Ref	Ref
Known HIV negative		4 (11.1)		11 (18.6)	1.21	0.61–2.43	0.586
Unknown		13 (36.1)		19 (32.2)	0.98	0.55–1.75	0.951
Caregiver's social support
Disclosure to spouse	36	34 (94.4)	58	52 (89.7)	0.81	0.35–1.88	0.622
Concerns for inadvertent disclosure of caregiver HIV status	50	27 (54.0)	75	31 (41.3)	0.80	0.50–1.29	0.360
Need spouse permission to test child	36	7 (19.4)	59	12 (20.3)	1.02	0.54–1.94	0.955
Perceived stigma at school or community for child	50	31 (62.0)	75	35 (46.7)	0.75	0.46–1.22	0.243
Concerns for child finding out being positive	50	32 (64.0)	75	40 (53.3)	0.80	0.49–1.33	0.396
Child perceived not being sickb	27	8 (29.6)	50	13 (26.0)	0.84	0.41–1.75	0.647
Economic factors
Has income	50	31 (62.0)	75	54 (72.0)	1.21	0.73–2.00	0.456
Monthly income (per 10 USD)b	31	$86 ($65–$234)	54	$78 ($65–$182)	0.99	0.97–1.02	0.706
Owns home	50	2 (4.0)	75	6 (8.0)	1.35	0.53–3.43	0.526
Monthly rent (per 10 USD)b	48	$32 ($19–$39)	69	$32 ($19–$52)	1.02	0.97–1.08	0.405
Time to travel to clinic (per 10 min)	50	30 (20–60)	75	45 (30–60)	1.00	0.97–1.04	0.809
Caregiver home distance to clinic (km)	24	2.9 (2.1–10.8)	29	10 (4.7–19.3)	1.01	0.96–1.06	0.752
Caregiver has cost to travel to clinic	50	32 (64.0)	75	44 (58.7)	0.91	0.57–1.45	0.697
Caregiver travel fare (per 1 USD)b	32	$0.71 ($0.58–$1.30)	44	$1.30 ($0.65–$1.95)	0.97	0.83–1.14	0.753
Children has travel cost to clinic	50	34 (68.0)	75	49 (65.3)	0.95	0.59–1.54	0.846
Travel cost if bring all children to clinic (per 1 USD)b	34	$1.6 ($0.78–$2.60)	49	$2.60 ($1.30–$4.68)	1.09	0.97–1.23	0.160
Miss work to test child	50	19 (38.0)	75	31 (41.3)	1.06	0.67–1.67	0.814
Has NHIFa card	50	13 (26.0)	75	32 (42.7)	1.33	0.84–2.12	0.226
Tested children at home (vs. at clinic)	50	14 (28.0)	75	34 (45.3)	1.33	0.84–2.10	0.216

^aNHIF card: National Health Insurance Fund card in Kenya

^bData collected in 2014, all monetary values are converted and inflated from 2014 Kenya shillings to 2020 US dollars values

^cRR adjusted for sites and clustered on family

Child-Level Factors Associated with Being Tested

Among families that tested some but not all children of unknown HIV status, we aimed to determine whether child characteristics were associated with whether a child was tested or not. We assessed 151 children from 50 caregivers who had more than one child of unknown status. Among these children, 66 (44%) were not tested and 85 (56%) were tested. The number of children in each family ranged from 2 to 6. Child demographics, HIV exposure or prevention history factors were not associated with the caregivers’ decision of whether or not to test a child in the same family (Table 4).

Table 4.

Child-level cofactors associated with caregiver’s decision of whether or not to test a child in a family among families in which at least one but not all children were tested in the study

Cofactors	Not tested (N = 66)		Tested (N = 85)		RRc	95% CI	p-value
	N	n (%) or median (IQR) or mean (SD)	N	n (%) or median (IQR) or mean (SD)
Child demographics
Female	66	35 (53.0)	85	50 (58.8)	1.12	0.73–1.74	0.598
Age (years)	66	7 (1–11)	85	7 (4–10)	1.02	0.97–1.08	0.429
Birth order (mean [SD])	66	Mean 2.1, SD 1.10	85	Mean 2.2, SD 1.13	1.03	0.85–1.24	0.768
Born at a facility (vs. at home)b Child's clinical history	37	33 (89.2)	47	35 (74.5)	0.68	0.35–1.31	0.244
Child's clinical history
Months breastfed	65	12 (6–24)	82	12 (9–20)	1.01	0.98–1.03	0.571
Child PMTCTa	38		52
Mother tested negative or not tested		20 (52.6)		41 (78.8)	Ref	Ref	Ref
Mother HIV positive and no EID testing		3 (7.9)		7.7	0.74	0.23–2.39	0.613
Mother HIV positive and child EID tested		15 (39.5)		8 (15.4)	9.52	0.24–1.11	0.093
Has been hospitalizedb	49	3 (6.1)	64	1 (1.6)	0.44	0.06–3.24	0.422

^aUnknown testing behaviors and status (n = 11; 9.9%) from female caregivers were not included for univariate analysis. Responses from male caregivers were excluded due to missing data

^bVariable collected after a certain date; missing data is likely at random

^cRR adjusted for sites and clustered on family

Discussion

In this study, we found that caregiver demographics and HIV history, but not child-level factors, economic, travel, or social support factors, were associated with selection of test location and which children to test. Among caregivers who tested any children, caregiver sex, spouse HIV status, and number of untested children were associated with choice of testing children at home or clinic. Among families who tested any children, having fewer children to test was associated with completing testing for all children.

Our findings suggest that HBT and CBT appeal to different groups and offering HBT may complement CBT as an alternative testing venue for index case pediatric testing. We found caregivers with more children of unknown HIV status preferred HBT and were less likely to complete testing for all of their children. This may be attributed to increased child care needs compared with those with fewer children, as studies from Kenya and South Africa have suggested that child care needs and responsibility for multiple children were barriers for pediatric HIV testing [19, 24]. By reducing efforts required to take children to clinic for HIV testing, HBT has potential to increase pediatric HIV testing coverage for some families.

We also found that caregivers with known HIV-negative spouses were twice as likely to test children at home compared to those with spouses known to be living with HIV or of unknown status; this effect was driven by female caregivers in our sample. This may be due to concerns of facing discrimination if community members saw that the child was taken to an HIV clinic [25]. Multiple qualitative studies have found that fear of stigma—both in the form of fear of accessing testing and fear of a positive diagnosis—is a barrier to testing children and adolescents [19, 24, 26-30], and that clinic testing can be an experience potentially associated with stigma [25]. HBT could be the more private alternative testing location and potentially a lower stigma choice for some caregivers.

We found that female caregivers were less likely than male caregivers to test at home. Men face more barriers to taking children to a clinic for testing due to gender norms commonly seen in the region—a Zambian study noted men’s reluctance to take children to test at clinic due to fear of losing reputation through being associated with HIV themselves [29]. On the other hand, women may also face loss of reputation with HBT and may want to avoid negotiations at home, where they hold less decision-making power compared to men in the household [29]. The findings from our study emphasize the need for qualitative evidence to gain deeper understanding of the intertwined relationship of stigma, decision-making power, gender inequality, and caregivers’ social interactions and their effects on decisions of pediatric HIV testing.

Our findings suggest that caregiver-level factors were more influential than child-level factors in pediatric HIV testing decisions. Caregiver sex, spouse HIV status, number of untested children were the most decisive factors in testing location preferences and testing uptake behaviors. These findings align with evidence from Uganda and Mozambique that compared facility or community factors to caregiver-level factors and found caregiver-level factors to be the most critical drivers in uptake of child HIV testing, including knowledge about HIV and prevention, fear of stigma, and social support [30, 31]. In contrast, we did not identify child-level demographic or HIV prevention history factors associated with caregivers’ testing decisions in our study. There is very little evidence on child-level testing predictors; however, one study in Cameroon found that older children were less likely to be tested for HIV in index case testing [32], potentially due to being less available because of school attendance.

Our study had limitations; we sampled 7 facilities in Kenya, which may limit generalizability. Correlates not found statistically significant, such as stigma, economic cost, distance to adult care clinic and child characteristics, might be true influential factors but were not able to stand out due to a relatively small sample size. Finally, the caregivers included in this study were mostly female and the male caregivers had more missing data reported. This study’s findings were weighted more towards the female caregivers’ voices and may not fully reflect male caregivers’ preferences and behaviors.

Conclusion

Caregiver-level factors were more influential than child-level factors in pediatric HIV testing decisions. HBT appears to be preferable to families with higher child-care needs and may be a more private and lower stigma choice to encourage pediatric HIV testing uptake than clinic testing for some families. The study findings call for integrated approaches in providing pediatric HIV testing services, such as incorporating and strengthening stigma-reduction and HIV care counseling, building skills in existing partner services, emphasizing the benefit of testing children, and providing HBT to accommodate diverse needs.