JOURNAL OF TROPICAL PEDIATRICS, VOL. 57, NO.

1, 2011

Age and CD4 count of vertically HIV-infected children at

the time of diagnosis: what are independent predictors for
being symptomatic and CD4 counts drop?
by Yifru Berhan
Department of Medicine, Hawassa University, PO Box 1560, Awassa, Ethiopia

Correspondence: Yifru Berhan, MD, Associate Professor, Hawassa University, College of Medical and Health Sciesnces,

Downloaded from https://academic.oup.com/tropej/article/57/1/14/1723764 by guest on 16 December 2020

PO Box 1560, Awassa, Ethiopia. E-mail: <[email protected]>; <[email protected]>.

Summary
A review of the literature has revealed that data on HIV-infected clinical presentations, age at the time
of diagnosis and level of immunosuppression in resource-poor settings are very limited. A multicenter
retrospective and cross-sectional method was used to analyze 1163 children <15 years of age. More
than half of the children were >5 years of age (mean  SD age 4.9  3.2). About 54% of children were
symptomatic. Tuberculosis and chronic dermatologic disorders were the commonest co-infections. The
severity of immunosuppression was highest in preschool children (46.6%) and early adolescents
(41.3%). After adjustment for sex, age, pattern of feeding and hemoglobin level, multinomial logistic
regression showed that CD4 count 200–499, 500–999 and Tigray ethnicity were independently
associated with being symptomatic. More than one-third of the children were in a state of severe
immunosuppression and more than half were immunologically eligible for antiretroviral treatment.

Key words: children, symptomatic, CD4 count, severe immunosuppression, Ethiopia.

Introduction of them being in Sub-Saharan Africa [1]. A prospect-

Every year about 600 000 children acquire HIV infec- ive cohort study done in Rwanda concluded that in
tion worldwide, mainly through vertical route, >90% Africa, vertically HIV-1-infected children develop
disease manifestations early in life [2]. Worldwide,
some authors pointed out that children accounted
for 20% of AIDS deaths, whereby rapid progression
Acknowledgements to disease was responsible for high mortality [3].
The author specially acknowledges The Netherlands According to them the occurrence of the infection
Government for supporting this project through its prior to birth, when the immune system is yet to
Program for the Institutional Strengthening in develop completely, is the major determinant for
Post-Secondary Education and Training Capacity rapid HIV disease progression. A recently published
(NPT/ETH 107), which was implemented in collab- article [4] stated that there is some evidence support-
oration with Maastricht University via its MUNDO ing rapid progression of the disease to AIDS
office. The author also thanks Mr Heinz Greijn for syndrome in resource-poor settings. However,
his friendly and effective coordination starting data on this area is very limited. The same study
from inception to conclusion phase. He would identified that child mortality was independently
like to extend his deep gratitude to team leaders of associated with maternal HIV status and death,
the data collectors Dr Desta (Mekele hospital resulting in an almost 4 fold increase in mortality
ART clinic), Dr Fozia and Sr Tizita (Gondar hos- by age 2 years [4].
pital ART clinic), Sr Haregewoin (Yekatit 12 hos- Another author described the natural course of
pital pediatric ART clinic), Sr Zerfe (Zewuditu vertically acquired HIV infection as bimodal unless
hospital pediatric ART clinic), Sr Misra (Jimma hos- put on antiretroviral therapy (ART). In 15–20% of
pital ART clinic), Ato Alemayehu (Awassa referral children there will be rapid progression with
hospital ART clinic) and Ato Tamirat (Yirgalem AIDS-defining illnesses occurring within the first
hospital ART clinic), and to study participants for year of life. In contrast, in 80–85% of the children
their unconditional involvement. there is much slower progress, which may take on

Published by Oxford University Press 2010. All rights reserved. For Permissions, please email: [email protected] 14
doi:10.1093/tropej/fmq032 Advance Access published on 19 May 2010
Y. BERHAN

average 8 years to manifest AIDS-defining symp- infection and to identify the predictors for being
toms [5]. A prospective study that involved 11 symptomatic, commencing ART and decline in
European centers showed progression of the disease CD4 counts.
to either category C or death by age 1 year in >15%
of the children vertically infected with HIV, and
nearly 50% by age 10 years, with fast progression Methods
among children who were not on ART. The same The data required for this study were collected partly
study revealed that after 4 years of age, <25% by reviewing the medical records and partly by inter-
become symptomatic irrespective of ART [6]. viewing parents or guardians of HIV-positive chil-
Another study has shown that 40% of vertically dren <15 years of age who had follow-up for ART
HIV-infected children progressed to a category C or chronic care (pre-ART) between August 2008 and
diagnosis by age 3 years and 51% of them died. March 2009 in seven referral public hospitals. These

Downloaded from https://academic.oup.com/tropej/article/57/1/14/1723764 by guest on 16 December 2020

Untreated children were significantly more likely to hospitals are located in different regional states of the
progress to a category C diagnosis (62% without Federal Democratic Republic of Ethiopia with differ-
ART vs. 28% on ART) [7]. ent distances from the capital city (Addis Ababa):
The clinical manifestations observed in adults Mekele (777 km North), Gondar (750 km North
during seroconversion are not so evident in infants West), Jimma (330 km South West), Hawassa
and toddlers. This is because >95% of pediatric HIV (275 km South), Yirgalem (315 km South) and
infection is acquired through vertical routes (during Addis Ababa (Yekatit 12 and Zewuditu hospitals).
pregnancy/labor and lactation) [8]. In one retrospect- Because of the varied geographic locations,
ive cohort study, nearly one-fourth of the children HIV-infected children case loads, and clinical service
and adolescents aged 9–16 years living with perinatal- beneficiary population ethnic and cultural diversifi-
ly aquired HIV infection were asymptomatic and had cation, HIV infected children who had follow-up in
a relatively good immune system [9]. Although there these hospitals were expected to give a fairly accurate
is considerable variability in the presenting manifest-
national sample representation.
ations, disease progression and extent of immuno-
The Multi-Centered Retrospective Cohort Study in
suppression, some of the AIDS-defining diseases in
HIV-Positive Children in Ethiopia sample database
children commonly cited were Pneumocystis carinii
(1241 children) was used as data resource. This data-
pneumonia, recurrent bacterial infections, HIV en-
base included study participants by taking the preva-
cephalopathy, wasting syndrome and cytomegalo-
lence of maximum HIV vertical transmission among
virus disease [10].
children with no intervention at any stage (45%). The
Age-adjusted CD4 count is used to categorize im-
munologic status of HIV-infected children as having 2005 Ethiopian Demographic and Health Survey
no evidence of, or of having moderate or severe im- (EDHS) [14] finding on the national prevalence of
munosuppression. Absolute CD4 counts (in cells/ml) traditionally performed uvulectomy or tonsillectomy
of 750–1499 and <750 for infants, 500–999 and <500 among women (42%) as population prevalence was a
for children 1–5 years of age and 200–499 and <200 second option for estimation. Secondly, taking into
for children 6–12 years of age are expressed as mod- account the June 2008 national report of ever en-
erate and severe immunosuppression, respectively for rolled HIV-infected children (15 929 cases), and
each age category [11]. assuming 2.0 for the design effect, 95% confidence
A retrospective study in India showed that the interval (CI), 0.04 margin of error and 10% refusal,
common clinical manifestations in children at pres- sample size for single population proportion was
entation were oral candidiasis, pulmonary tubercu- determined using EpiInfo computer software (version
losis, recurrent respiratory tract infections, 2002) [15].
papulo-pruritic dermatitis, hepatosplenomegaly, For this study, 1163 HIV-infected children from
lymphadenopathy and chronic diarrhea [12]. the database were eligible for analysis. Children
In Ethiopia, despite being one of the Sub-Saharan who were diagnosed to have HIV infection other
African countries highly affected by HIV infection than the study hospitals and transferred without a
[13], to the best of the author’s knowledge, there is complete data set were excluded. Additionally, even
no published study that has assessed the clinical pres- in the study hospitals, failure to extract important
entations, age at the time of diagnosis, correlation of information for this study in either the medical
age with CD4 count in children who have been ver- charts or electronic records led to exclusion from
tically infected with HIV. The author hypothesized the analysis.
that the survival of HIV-infected children beyond the A retrospective review of medical chart or electron-
age of 5 years without ART would have significantly ic records version of HIV-infected children enabled
decreased and thus, more than three-fourths of the us to collect the baseline data at the time of diagnosis.
study participants would be preschool children. The This included the socio-demographic variables/iden-
objective of this study was to describe the age and tifiers, absolute CD4 count, hemoglobin level, date
CD4 level of children at the time of diagnosis of HIV ART started, reasons why they came to hospital and

Journal of Tropical Pediatrics Vol. 57, No. 1 15

 Y. BERHAN

clinical problems identified, pattern of breast feeding 200–499, 500–999 and 1000. The raw CD4 data
and the health status of the parents. were used to correlate with raw age data.
The principal research protocol initially received HIV-infected children were evaluated at the first
ethical clearance from the University’s Institutional visit and during follow-up usually by general practi-
Review Board. Subsequently, an official letter was tioner physicians and rarely by pediatricians.
submitted for approval to each study hospital’s med- Microsoft office Excel 2003 and SPSS version 16.0
ical director prior to data collection. A statement and computer software were used to analyze the data.
certificate of consent was attached to each question- Multinomial logistic regression was performed on
naire to let the parents or guardian of the child read data for clinically symptomatic children, for those
or was read to them by data collectors. Interviews started on ART and for CD4 cell counts
were conducted after the child’s caretaker had given <500 mm3. P-value <0.05 was taken as a statistic-
written informed consent. Furthermore, confidential- ally significant association.

Downloaded from https://academic.oup.com/tropej/article/57/1/14/1723764 by guest on 16 December 2020

ity and anonymity was assured by analyzing and
disseminating the study findings in aggregate.
In describing this study, the following definitions Results
are used. Infant: <12 months of age; preschool chil- Of 1163 HIV-positive children who had follow-up in
dren: 1–4.9 years of age; early school children: the study hospitals during the 8-month study period,
5–9.9 years of age; early adolescent: 10–14.9 years 640 (55.0%) were girls. The three most common
of age. Symptomatic: children who came to the en- ethnic groups were Amhara 502 (43.2%), Tigray
rolling hospital for another medical problem and as 259 (22.3%) and Oromo 238 (20.5%). A total of
well diagnosed to have HIV infection. 1150 (98.9%) mothers were seropositive for HIV in-
Asymptomatic: children who came to enrolling hos- fection and the remaining 1.1% mothers’ serostatus
pital with no history of illness that caused concern to was unknown.
parents/guardian or required health professional or As presented in Fig. 1, the age of children when
traditional healers’ intervention when they were diag- diagnosed with HIV infection was in the range of
nosed with HIV infection. 1 month to 14 years with median and mean  SD
As immunologic marker, absolute CD4 count per ages 5 and 4.9  3.2 years, respectively. A total of
cubic millimeter is used for analysis. In all the study 576 (49.5%) children were diagnosed before the age
hospitals, CD4 cell count was determined by flow of 5 years, 480 (41.3%) 5–9.9 years and 107 (9.2%)
cytometry and always in the morning. The absolute 10–14.9 years. As shown in Fig. 2, the clinical pres-
CD4 cells count per cubic millimeter of children at entation at the time of diagnosis in relation to the
the time of diagnosis is grossly grouped into <200, chronologic age showed that more than half of the

180
160
160

140 128
117 115
Number of children

120 106 110 108

100 90

80 65
57
60 48

40 28
19
20 9
3
0
.9

.9

.9

.9
.9

.9

.9

.9
1

.9

11 9

12 9

13 9

14 9
4.9
3.
0.

1.

2.
-6

-7

-8

-9
-1

-2

-3

-4

-5
<

-1
-1

-1

-1

-1
6

9
1

10

Age in years, HIV diagnosed

FIG. 1. Age of children when diagnosed to have HIV infection, Ethiopia 2008/09. N ¼ 1163; mean  SD: 4.9  3.2
years; median age: 5 years.

16 Journal of Tropical Pediatrics Vol. 57, No. 1

Y. BERHAN

350

286(46%)
300
261(49%)
250(40%)
250

Number of children
194(36%)
200

150

100
39(6%) 52(8%) 55(10%)

Downloaded from https://academic.oup.com/tropej/article/57/1/14/1723764 by guest on 16 December 2020

50 26(5%)

0
<1 1 - 4.9 5 - 9.9 10 - 14.9
Age in years
Symptomatic (N = 627) Asymptomatic (N = 536)

FIG. 2. Pattern of presentation of children in relation to their chronologic age at the time of diagnosis for HIV
infection, Ethiopia 2008/09.

TABLE 1
The CD4 cell counts of HIV-infected children in different age category in no. (%), Ethiopia 2008/09

Age (years)a Total (n) CD4 cell count mm3

<200 200–499 500–999 1000

<1 65 4 (6.2) 19 (29.2) 17 (26.2) 25 (38.4)

1–4.9 511 63 (12.3) 176 (34.5) 179 (35.0) 93 (18.2)
5–9.9 480 106 (22.1) 195 (40.6) 135 (28.1) 44 (9.2)
10–14.9 107 44 (41.1) 37 (34.6) 19 (17.8) 7 (6.5)
Total 1163 217 (18.7) 427 (36.7) 350 (30.1) 169 (14.5)
a
Age of children at the time of diagnosis of HIV infection.

TABLE 2
children were brought to the hospitals because they The relationship of CD4 count and clinical manifest-
were symptomatically sick, 627 (53.9%) of 1163. ations in HIV-infected children, Ethiopia 2008/09
Although there was a linear increment in the
number of sick children against their age up to 10 CD4 Total Symptomatic Asymptomatic
years, the difference in percentage between symptom- (cells mm3) (n) no. (%) no. (%)
atic and asymptomatic children in each age category
was not statistically significant. <200 217 109 (50.2) 108 (49.8)
200–499 427 244 (57.1) 183 (42.9)
Among the symptomatic group, the top five clin-
500–999 350 196 (56.0) 154 (44.0)
ical problems why children came to the hospitals at 1000 169 78 (46.2) 91 (53.8)
the time of diagnosis for HIV infection were tuber-
culosis 182 (29.0%), chronic dermatologic disorders
92 (14.7%), recurrent diarrheal diseases 84 (13.4%),
recurrent pneumonia 56 (8.9%) and herpes zoster 39
(6.2%). Of 636 children diagnosed to have HIV in- accounted for 217 (18.7%), 427 (36.7%), 350
fection while being asymptomatic, more than (30.1%) and 169 (14.5%), respectively (Table 1). Of
three-fifths (63.1%) were subjected to a HIV test be- the 217 children with CD4 count <200 mm3,
cause either or both the parents knew that they were 67 (31%) and 150 (69%) were below and above
living with HIV infection. 5 years of age, respectively.
At the time of diagnosis of HIV infection, 644 The CD4 count of 95 (8.2%) children was
(55.4%) children had CD4 counts <500 mm3. <100 mm3. Cross tabulation of clinical features
Specifically, it was found that children with CD4 with CD4 count (Table 2) showed similar finding as
counts <200, 200–499, 500–999 and 1000 mm3 shown in Fig. 2. The proportion of symptomatic

Journal of Tropical Pediatrics Vol. 57, No. 1 17

 Y. BERHAN

Downloaded from https://academic.oup.com/tropej/article/57/1/14/1723764 by guest on 16 December 2020

FIG. 3. Age, CD4 count and clinical condition of children at the time of diagnosis for HIV infection, Ethiopia
2008/09. The regression lines show the correlation between age and CD4 count as paneled to clinical manifest-
ations. Pearson correlation coefficient for asymptomatic (r1) ¼0.45 (p ¼ 0.01); symptomatic (r2) ¼ 0.4
(p ¼ 0.01).

children with CD4 count <200, 200–499, 500–999 shown in Fig. 4. The ratio of children on ART to
and 1000 mm3 were 109 (50.2%), 244 (57.1%), pre-ART was about equal between 1 and
196 (56.0%) and 78 (46.2%), respectively. 10 years of age, whereas 67 (62.6%) of early
In Fig. 3, the variation in CD4 counts with age and adolescents were on ART. The finding that the
clinical condition of children at the time of diagnosis majority of CD4 counts at the time of diagno-
for HIV infection is presented. In both symptomatic sis being between 0 and 500 cells per cubic milli-
and asymptomatic children at the time of diagnosis meter in those children who were put on ART
there is negative correlation between CD4 count and indicates that the CD4 count was dropping faster
chronologic age [Pearson correlation coefficient for among untreated children as they grew older.
asymptomatic (r1) ¼ 0.45 (p < 0.01) and for symp- Although the regression lines in both groups show
tomatic (r2) ¼ 0.4 (p < 0.01)]. negative correlations as children grew older, it is
In Table 3, multinomial logistic regression was marked in pre-ART group (r1 ¼ 0.25 vs.
done for variables thought to have association with r2 ¼ 0.51).
children’s illness (symptomatic) at the time of diag- As presented in Table 4, the unadjusted logis-
nosis for HIV infection. It has been shown that tic regression showed significant association with
children of Tigray ethnicity [p ¼ 0.01, adjusted odds male (p ¼ 0.007; crude OR 1.4; 95% CI 1.09–1.74),
ratio (OR) 0.59, 95% CI 0.39–0.89], with CD4 counts age 10–14.9 years (p ¼ 0.036; crude OR 2.0; 95%
200–499 mm3 ( p ¼ 0.028, adjusted OR 1.5, 95% CI CI 1.05–3.65), mother dead (p ¼ 0.01; crude OR 1.5;
1.05–2.20) and 500–999 mm3 (p ¼ 0.032, adjusted 95% CI 1.1–2.0), CD4 count <1000 mm3 and
OR 1.5, 95% CI 1.04–2.22) were significantly more hemoglobin between 7 and 11.5 g dl1. After adjust-
symptomatic compared to those who were asymp- ment for sex, mother loss, father loss and hemo-
tomatic, after adjustment for sex, age, pattern of globin level, however, only age and CD4 count
feeding in the first 6 months of infancy and hemoglo- independently predicted initiation of ART at
bin level at the time of enrollment. the time of enrollment, with adjusted OR 0.47 (age
Age, CD4 count at the time of diagnosis and 5–9.9 years), 0.42 (age 10–14.9 years), 336.9 (CD4
modes of treatment set (ART or pre-ART) are <200 mm3) and 8.9 (CD4 200–499 mm3).

18 Journal of Tropical Pediatrics Vol. 57, No. 1

Y. BERHAN

TABLE 3
Multinomial logistic regression for clinically symptomatic children at the time of diagnosis for HIV infection,
Ethiopia 2008/09

Variables Total (n) Symptomatic Crude Adjusteda

p-value OR (95% CI) p-value OR (95% CI)

Sex
Boys 523 297 (56.8) 0.075 1.2 (0.98, 1.56) 0.1 0.8 (0.65, 1.04)
Girls 640 330 (51.6) 1 1
Age (years)b
<1 65 39 (60.0) 1 1
1–4.9 511 250 (48.9) 0.09 0.64 (0.38, 1.08) 0.13 0.65 (0.37, 1.13)

Downloaded from https://academic.oup.com/tropej/article/57/1/14/1723764 by guest on 16 December 2020

5–9.9 480 286 (59.6) 0.95 0.98 (0.58, 1.67) 0.88 0.96 (0.54, 1.7)
10–14.9 107 52 (48.6) 0.15 0.63 (0.34, 1.18) 0.14 0.60 (0.31, 1.18)
Ethnicity
Amhara 502 279 (55.6) 0.36 0.85 (0.59, 1.21) 0.46 0.87 (0.60, 1.26)
Oromo 238 133 (55.9) 0.44 0.85 (0.57, 1.28) 0.62 0.90 (0.59, 1.36)
Tigray 259 117 (45.2) 0.003 0.55 (0.37, 0.82) 0.01 0.59 (0.39, 0.89)
Other 164 98 (59.8) 1 1
Mode child fedc
Breast milk 540 298 (55.2) 1 1
Mixed 570 298 (52.3) 0.64 0.87 (0.49, 1.55) 0.76 0.9 (0.49, 1.7)
Replacement 53 31 (58.5) 0.39 0.78 (0.44, 1.38) 0.47 0.8 (0.44, 1.5)
CD4 count (mm3)
<200 217 109 (50.2) 0.427 1.2 (0.79, 1.76) 0.67 1.1 (0.72, 1.68)
200–499 427 244 (57.1) 0.016 1.6 (1.09, 2.23) 0.028 1.5 (1.05, 2.20)
500–999 350 196 (56.0) 0.036 1.5 (1.03, 2.15) 0.032 1.5 (1.04, 2.22)
1000 169 78 (46.2) 1 1
Hemoglobin (g dl1)
<7 49 22 (44.9) 0.33 0.75 (0.42, 1.34) 0.36 0.75 (0.41, 1.37)
7–9.9 213 115 (54.0) 0.66 1.07 (0.78, 1.47) 0.75 1.06 (0.76, 1.47)
10–11.5 317 185 (58.4) 0.08 1.28 (0.97, 1.69) 0.10 1.27 (0.95, 1.69)
>11.5 584 305 (52.2) 1 1
a
Adjusted for all variables in this table.
b
Age of children at the time of diagnosis of HIV infection.
c
In the first 6 months: breast milk—exclusive breast milk; mixed—breast milk and any additional intake; replacement—
exclusive replacement feeding.

Multinomial logistic regression was also separately separately for the ART and pre-ART groups. For
done for CD4 cell counts <200 and 200–499 mm3 as children on ART, the level of CD4 at the time of
dependent variables and sex, age at the time of diag- enrollment was <200 mm3 for 66% of early adoles-
nosis, ethnicity, residence, hemoglobin level and pres- cent age, 43% of early school age children and 23%
ence of co-infections as independent variables (data of preschoolers. Majority of preschool and early
not shown). After adjustment, children’s age 1 year school children were found to have CD4 counts in
was independently associated with both CD4 <200 the range of 200–499 mm3. Among children who
and 200–499 mm3: for CD4 <200 (age 1–4.9 years, were not eligible for ART at the time of presentation,
p ¼ 0.007, adjusted OR 4.6, 95% CI 1.5–14.1; age more than half of early adolescent age were found to
5–9.9 years, p < 0.0001, adjusted OR 17.4, 95% CI have CD4 counts in the range of 200–499 mm3,
5.6–53.9; age 10–14.9 years, p < 0.0001, adjusted OR while among the <10-year-olds, CD4 counts were
43.6, 95% CI 11.4–167.0). The independent associ- mainly >500 cells mm3.
ation of CD4 count with children’s age was very The CD4 count of 414 (35.6%) children was in the
marked in 10 years (p < 0.0001, adjusted OR 7.3, range of severe immunosuppression. In detail, 46.8%
95% CI 2.6–20.4). Being symptomatic at the time of (239/511) of preschool children, 22.1% (106/480) of
diagnosis was also an independent predictor of CD4 early school children and 41.3% (44/107) of early
count 200–499 mm3 (p < 0.05, adjusted OR 1.5, adolescents were identified to have severe immuno-
95% CI 1.1–2.2). suppression. On the contrary, 318 (27.3%) of the
Figure 5 shows the percentage of preschool, early total children (17.5% age 5 years) were found to
school and early adolescent children in relation to have normal CD4 counts. Overall, 147 (12.6%) chil-
their absolute CD4 count at the time of diagnosis, dren were asymptomatic and CD4 counts were

Journal of Tropical Pediatrics Vol. 57, No. 1 19

 Y. BERHAN

Downloaded from https://academic.oup.com/tropej/article/57/1/14/1723764 by guest on 16 December 2020

FIG. 4. Age, CD4 count and use of ART at the time of diagnosis for HIV infection, Ethiopia 2008/09. The
regression lines show the correlation between age and CD4 count as paneled to kind of therapy. Pearson cor-
relation coefficient for children on ART (r1) ¼ 0.33 (p ¼ 0.01); pre-ART (r2) ¼ 0.36 (p ¼ 0.01).

TABLE 4
Multinomial logistic regression of children with HIV for potential factors for initiation of ART, Ethiopia 2008/09

Variables Total (n) On ART Crude Adjusteda

p-value OR (95% CI) p-value OR (95% CI)

Sex
Boys 523 294 (56.2) 0.007 1.4 (1.09, 1.74) 0.067 0.77 (0.58, 1.02)
Girls 640 309 (48.3) 1 1
Age (years)b
<1 65 30 (46.2) 1 1
1–4.9 511 261 (51.1) 0.455 1.2 (0.73, 2.0) 0.56 0.83 (0.45, 1.53)
5–9.9 480 245 (51.0) 0.46 1.2 (0.72, 2.05) 0.019 0.47 (0.25, 0.88)
10–14.9 107 67 (62.6) 0.036 2.0 (1.05, 3.65) 0.033 0.42 (0.19, 0.93)
Mother alive
Yes 938 439 (46.8) 1 1
No 225 134 (59.6) 0.01 1.5 (1.1, 2.0) 0.27 1.2 (0.85, 1.8)
Father alive
Yes 748 384 (51.3) 1 1
No 415 219 (52.8) 0.64 1.1 (0.83, 1.35) 0.62 0.93 (0.69, 1.25)
CD4 count (mm3)
<200 217 213 (98.2) <0.0001 246.7 (85.1, 715.7) <0.0001 336.9 (113.7, 998.1)
200–499 427 261 (61.1) <0.0001 7.3 (4.69, 11.31) <0.0001 8.9 (5.6, 14.2)
500–999 350 99 (28.3) <0.0001 1.8 (1.16, 2.89) 0.004 2.0 (1.3, 3.2)
1000 169 30 (17.8) 1 1
Hemoglobin (g dl1)
<7 49 28 (57.1) 0.186 1.5 (0.83, 2.68) 0.22 1.6 (0.77, 3.2)
7–9.9 213 126 (59.2) 0.003 1.6 (1.2, 2.2) 0.057 1.5 (0.99, 2.2)
10–11.5 317 173 (54.6) 0.036 1.3 (1.02, 1.76) 0.32 1.2 (0.85, 1.7)
>11.5 584 276 (47.3) 1

n ¼ Pre-ART þ ART.
a
Adjusted for all variables in this table.
b
Age of children at the time of diagnosis of HIV infection.
20 Journal of Tropical Pediatrics Vol. 57, No. 1
Y. BERHAN

A 70 120
66%
60 100 98%
50 47%
80 82%
72%

Percent
40 43% 46%
60 61%
30
24% 40
20 23% 21% 39%
28%
10 20 18%
9% 7%
4% 2%
0 1% 0
CD4 < 200 CD4 200 - 499 CD4 500 - 999 CD4 1000+
B 60

Downloaded from https://academic.oup.com/tropej/article/57/1/14/1723764 by guest on 16 December 2020

57% On ART (N = 603) Pre-ART (N = 560)
50 48%
47%
40 FIG. 6. The aggregate distribution of absolute CD4
34% counts among children put on ART and chronic care
30 33% 30%
22%
(pre-ART), Ethiopia 2008/09.
20
18%
10 10%

0 10
Discussion
< 200 200 - 499 500 - 999 1000+ As a previous study showed [16] and parents of chil-
CD4 count
dren in this study believed, >95% of children acquire
HIV infection from their mother either before birth
or from breastfeeding. Despite this, probably because
of parents’ or guardians’ unawareness or ignorance,
FIG. 5. The percentage of preschool, early school >50% of the children were diagnosed with HIV in-
and early adolescent children in relation to their ab- fection after the age 5 years, and >9% not until early
solute CD4 count at the time of diagnosis, Ethiopia adolescence age. The mean age of the children at the
2008/09. (A) Line graph for children started on ART time of diagnosis was about 5 years which is earlier
at the time of diagnosis (first graph). (B) Line graph than a report from Uganda [17] but later than a
for children kept on pre-ART care (second graph). report from South Africa [18].
Much of the delay in making diagnosis for HIV
infection is very likely disadvantageous, particularly
within normal range. Specifically, 7 (10.8%) infants, for those children who were potentially eligible for
59 (11.6%) preschool children, 66 (13.8%) early ART. This is because many studies have shown that
school children and 15 (14.0%) early adolescents early diagnosis and timely commencement of ART
were asymptomatic and their CD4 counts were markedly improves the survival of children living
>1500, >1000 and > 500 mm3, respectively. with HIV infection [19, 20]. The apparent delay in
As presented in Table 4, of 65 infants in total, 30 HIV diagnosis in this study is also indirect evidence
were on ART. Their CD4 count distribution is not of low test uptake. Despite freely accessible nation-
shown in Fig. 5 because of marked difference in clas- wide HIV testing, the finding that 53.9% children
sification of immunosuppression in children. Among came to the hospitals primarily for serious and recur-
infants put on ART, 20, 4 and 6 were found to have rent illnesses is another proxy indicator of the reluc-
CD4 counts <750, 750–1499 and >1500 mm3, re- tance to have a serological test for HIV infection.
spectively. Similarly, 7, 9 and 19 children of Among the illnesses that urged parents or guard-
pre-ART group were with CD4 count <750, ians to let take the children to hospitals, pulmonary
750–1499 and >1500 mm3, respectively. The aggre- or extra-pulmonary tuberculosis infections were
gate distribution of CD4 cell counts in percentage more than twice as common as other co-infections/
among children put on ART vs. chronic care disorders, probably because of its high prevalence in
(pre-ART) forms an ‘X’-lines graph with negatively the community [21]. In addition, the prevalence of
sloping ART group (Fig. 6). herpes zoster seems much higher than has been usu-
Parents or guardians were asked about how the ally observed in daily clinical practice. However, the
children acquired HIV infection. A total of 519 rate is similar to other reports of herpes zoster in
(44.6%) thought that the children acquired the patients receiving highly active ART and where rou-
virus from their mother during breastfeeding, 510 tine varicella zoster virus immunization is widely
(43.8%) during delivery or during breast feeding practiced [22]. Otherwise, the commonest clinical
and 60 (5.2%) in utero. Seventy four (6.4%) parents manifestations identified in this study are consistent
or guardian had no idea how the children acquired with the finding in India but markedly different from
the virus. a report from New Jersey [12].

Journal of Tropical Pediatrics Vol. 57, No. 1 21

 Y. BERHAN

On the other hand, there was no marked difference However, the prolonged delay in starting ART, in
in age between symptomatic and asymptomatic chil- particular for infants, is beyond the findings of an-
dren at the time of enrollment, 42.4% children be- other study in South Africa [18]. Giving detailed ana-
tween age 4.9 and 15 years were asymptomatic when lysis and explanation for failure to start ART is
the HIV diagnosis was made, which may indicate the beyond the remit of this article. Nevertheless, since
possibility of some prolonged survival long even there was high prevalence of tuberculosis infection
without ART as reported in other studies [9]. among these study participants, it was initially enter-
Even if being asymptomatic was not associated tained as a possible factor for the delay in initiating
with age as shown in Fig. 2 and logistic regressions, ART. However, cross tabulation did not demon-
the overall drop in CD4 counts in both the symptom- strate association.
atic and asymptomatic groups as the age of children In conclusion, at the time of diagnosis, more than
advances is a strong immunologic suggestion of HIV one-third of the children were in a state of severe

Downloaded from https://academic.oup.com/tropej/article/57/1/14/1723764 by guest on 16 December 2020

disease progression. This finding was also analyzed in immunosuppression and more than half were im-
a logistic regression taking CD4 count as dependent munologically eligible for ART. Being clinically
variable and age and clinical presentations as inde- symptomatic at the time of diagnosis for HIV infec-
pendent variables at the time of diagnosis. The latter tion was found to have independent association with
two were independently associated with a drop in Tigray ethnicity and CD4 cell counts <500 mm3,
CD4 count. after adjustment for sex, age, pattern of feeding in
Generally, although about 46% of children were the first 6 months of infancy and hemoglobin level.
clinically asymptomatic at the time of diagnosis, ac- There was negative correlation between CD4 count
cording to the CDC immunologic classification of and age regardless of clinical presentation and eligi-
HIV-infected children using absolute CD4 count bility for ART. This was further verified by logistic
[11], 35.6% were in state of severe immunosuppres- regression that demonstrated advancing age as an
sion. The degree of immunosuppression did not show independent predictor of CD4 cell count decline
a trend with age. The highest proportion of severe and eligibility for ART. This study showed that
immunosuppression was seen in preschool children there was substantial delay in HIV test and ART
and early adolescents. This may be because the uptake. As a result, a significant number of children
viral load is low early in infancy and progressively presented with marked immunologic suppression.
increases up to the age 5 years, and thereafter slowly These findings highlight the importance of early
decreases as the HIV specific immunity develops, screening of all children born to mothers with
which may last till late childhood [8]. However, the known HIV infection or who are of unknown
proportion of children in each age category with no serostatus.
immunologic evidence of immunosuppression
showed a slight and progressive increase with age, References
from 10.8% in infants and to 14.0% in early adoles- 1. Mofenson LM, Munderi P. Safety of antiretroviral pro-
cents, which was an unexpected finding. In summary, phylaxis of perinatal transmission of HIV infected preg-
of 1163 children living with HIV, 12.6% were asymp- nant women and their children. J Acquir Immune Defic
Syndr 2002;30:200–15.
tomatic and their CD4 counts were within the CDC
2. Spira R, Lepage P, Msellati P, et al. Natural history of
normal range of reference values for each age cat- human immunodeficiency virus type 1 infection in chil-
egory. This finding may indicate that a number of dren: a five-year prospective study in Rwanda. Mother-
vertically HIV-infected children are slow progressors, to-Child HIV-1 Transmission Study Group. Pediatrics
which warrants further investigation. 1999;104:e56.
According to the Guideline for Pediatric HIV/ 3. Goulder PJ, Jeena P, Tudor-Williams G, et al. Pediatric
AIDS Care and Treatment in Ethiopia, the immuno- HIV infection: correlates of protective immunity and
logic criteria for starting ART in children using ab- global perspectives in prevention and management. Br
Med Bull 2001;58:89–108.
solute CD4 count are stated as: all infants regardless
4. Little K, Thorne C, Luo C, et al. Disease progression in
of CD4 count; CD4 count in cubic millimeter <750 children with vertically-acquired HIV infection in Sub-
for age 12–35 months; <350 for age 36–59 months Saharan Africa: reviewing the need for HIV treatment.
and <200 for 5 years [23].With these criteria, 684 Curr HIV Res 2007;5:139–53.
children were eligible, but only 593 were put on ART. 5. Domachowske JB. Pediatric human immunodeficiency
Breakdown by age showed that 40% (35/65) of in- virus infection. Clin Microbiol Rev 1996;9:448–68.
fants, 12.8% (44/343) of children 12–35 months and 6. Gray L, Newell ML, Thorne C, et al. Fluctuation in
7.1% (12/168) of children 36–59 months were not symptoms in human immunodeficiency virus-infected
children: the first 10 years of life. Pediatrics
started on ART during the period of enrollment
2001;108:116–22.
(till 6 months from the day of diagnosis). Usually, 7. Berk DR, Falkovitz-Halpern MS, Hill DW, et al.
commencing ART is not an emergency—there is Temporal trends in early clinical manifestations of peri-
time for a period of observation until readiness for natal HIV infection in a population-based cohort.
adherence with treatment can be ratified. JAMA 2005;293:2221–31.

22 Journal of Tropical Pediatrics Vol. 57, No. 1

Y. BERHAN

8. Niehues T, Lyall H. Antiretroviral therapy in children. 17. Mark MK, Kektinwa A, Maganda A, et al. Growth in
In: Rockstroh-Kamps Hoffmann (ed.). Paris, Cagliari, HIV-infected children receiving antiretroviral therapy
Wuppertal: HIV Medicine. Flying Publisher, 2006; at a pediatric infectious diseases clinic in Uganda.
pp. 375–9. AIDS Patient care STDS 2008;22:245–51.
9. Hernandez M, George DM, Laura GH, et al. Older 18. Feucht UD, Kinzer M, Kruger M. Reasons for delay in
children and adolescents living with perinatally initiation of antiretroviral therapy in a population of
acquired HIV infection. Pediatrics 1995;95:657–63. HIV-infected South African children. J Trop Pediatr
10. Ferris MG, Gordon ES. Natural history and classifica- 2007;53:398–402.
tion of pediatric HIV infection. UpToDate Desk top 19. Reddi A, Leeper SC, Grobler AC, et al. Preliminary
17.1. 2009. outcomes of a paediatric highly active antiretroviral
11. Centers for Disease Control and Prevention (CDC). therapy cohort from KwaZulu-Natal, South Africa.
Revised classification system of HIV Infection in chil- BMC Pediatr 2007;7:13.
dren <13 years of age. MMWR Recomm Rep 20. Zanchetta M, Ansemi A, Vendrame D, et al. Early

Downloaded from https://academic.oup.com/tropej/article/57/1/14/1723764 by guest on 16 December 2020

1994;43:1. therapy in HIV-1 infected children: effect on HIV-1
12. Madhivanan P, Mothi SN, Kumarasamy N, et al. dynamics and HIV-1 immune response. Antivir Ther
Clinical manifestations of HIV infected children. 2008;13:47–55.
Indian J Pediatr 2003;70:615–20. 21. Bruchfeld J, Aderaye G, Palme IB, et al. Molecular
13. Joint United Nations Programme on HIV/AIDS epidemiology and drug resistance of Mycobacterium
(UNAIDS). 2008 Report on the global AIDS epidemic. tuberculosis isolates from Ethiopian pulmonary tuber-
http://www.unaids.org (June 2008, date last accessed). culosis patients with and without HIV infection. J Clin
14. Ethiopia Demographic and Health Survey 2005. Microbiol 2002;5:1636–43.
16.11:252–6. 22. Sarah MW, Samir SS, Andrew PS, et al. Primary var-
15. Bland M. An Introduction to Medical Statistics icella and herpes zoster among HIV-infected children
3rd edn. Oxford University Press, 2000; pp. 335–46. from 1989 to 2006. Pediatrics 2008;121:e150–6.
16. Ferris MG, Kline M, Torchia MM, et al. 23. Federal HIV/AIDS Prevention and Control Office.
UpToDate version 17.1. Epidemiology of pediatric Guideline for Paediatric HIV/AIDS Care and
HIV infection. Treatment in Ethiopia. July 2008.

Journal of Tropical Pediatrics Vol. 57, No. 1 23