Risk Factors for Measles in HIV-Infected Children and Adolescents in Botswana

Abstract

We conducted a matched case-control study of 566 HIV-infected children in Botswana during a 2009-2010 measles outbreak to identify risk factors for measles. Children in the oldest age quartile (≥13.1 years) were four-fold more likely to acquire measles than those in the youngest quartile (<7.1 years). HIV-infected older children and adolescents may benefit from additional measles vaccination.

Introduction

Measles is a highly infectious viral disease that continues to be a major cause of global pediatric mortality and morbidity. Children in Sub-Saharan Africa may be particularly vulnerable due to malnutrition, suboptimal healthcare and HIV infection. Measles-related mortality is primarily due to an increased susceptibility to secondary bacterial and viral infections (1) resulting from direct mucosal damage by measles virus infection and measles-induced immune suppression (2). HIV-infected children are at higher risk of measles-related complications (3), in part due to their pre-existing immunosuppression.

HIV-infected children are at particular risk of developing measles because they receive fewer maternal measles antibodies in utero (4) and have high rates of both primary vaccine failure (when initial antibodies are insufficient to provide protection against disease) and secondary vaccine failure (when immunity wanes faster than normally expected) (5). Anti-retroviral therapy has become increasingly available allowing many more HIV-infected children to survive into adulthood, well beyond their standard measles immunization at nine months of age. It is important to study the HIV and measles co-infected population because they are at great risk of harm themselves and may play a role in decreasing herd immunity within generalized HIV epidemic settings.

Sub-Saharan Africa experienced a widespread measles epidemic between 2009 and 2010, with Botswana reporting 1,965 cases (6). We aimed to study the risk factors for measles in HIV-infected children and adolescents receiving care at the Botswana Baylor Children's Clinical Center of Excellence (BBCCCOE) during this epidemic. Botswana exhibited a unique combination of high HIV prevalence, sophisticated HIV care and a publicly funded vaccine program that at the time administered a single measles vaccine at nine months of age. BBCCCOE is one of the largest pediatric HIV providers in the country and maintains electronic medical records for more than 4,000 patients.

Materials and Methods

We conducted a matched case-control study among HIV-infected children and adolescents receiving outpatient HIV services at the BBCCCOE from October 2008 to October 2011. Data on potential risk factors for measles were collected retrospectively through a review of electronic medical records. This study was approved by the Institutional Review Boards of Botswana's Ministry of Health, Princess Marina Hospital, University of Botswana School of Medicine, Baylor College of Medicine, University of Pennsylvania, and Harvard School of Public Health.

Selection of Case and Control Subjects

Measles cases were identified using the clinical case definition recommended by the World Health Organization (WHO) for measles epidemics. Cases included patients in whom a clinician suspected measles or those with fever, maculopapular rash and either cough, coryza or conjunctivitis. Initially, we searched all electronic medical records at the BBCCCOE for the terms “measles,” and “koplik.” We then reviewed the Isolation Ward Registers at Princess Marina Hospital in Gaborone and Bamalete Lutheran Hospital in the village of Ramotswa for measles diagnoses. Two investigators (CJ, MC) assessed suspected cases for inclusion. All cases were then reviewed by a third investigator (ERW) to ensure each met the case definition. Case subjects included HIV-infected patients aged ≤18 years of age seen at the BBCCCOE at least once during the study period with a diagnosis of measles. Two control subjects were matched to each case based on the date of measles diagnosis (±2 days). Eligible controls were HIV-infected patients aged ≤18 years of age seen at the BBCCCOE at least once during the study period who were not diagnosed with measles.

Risk Factor Assessment

We collected information on the following covariates as potential risk factors for measles: age, sex, nutritional status, WHO HIV disease stage, absolute and percent CD4 cell count, HIV viral load, type and duration of antiretroviral therapy and receipt of Pneumocystis pneumonia (PCP) prophylaxis (trimethoprim/sulfamethoxazole). We classified children as moderately or severely malnourished using weight-for-height less than −2 standard deviations (SD) compared to a WHO reference population. WHO age-specific thresholds of CD4 cells (absolute and percent) were used to classify patients into categories of immunosuppression. HIV viral load was categorized as detectable if > 400 copies/mL. Type of antiretroviral therapy (first line, second line or greater, suspended or not yet on treatment) was categorized by the treating physician. We assessed all risk factors in relation to the date of the last clinic visit prior to the date of measles diagnosis in the case subject.

Statistical Analysis

To identify risk factors for measles, we used a two-stage approach based on univariable and multivariable conditional logistic regression models. First we assessed each risk factor individually; any factor with P≥0.20 was excluded from further consideration. The remaining covariates were then used to construct the multivariable model. In all models we conditioned on the matched set of case and control subjects defined by date of measles diagnosis for the case. All statistical analyses were conducted using SAS software version 9.3 (SAS Institute, Cary, NC).

Results

We identified 189 measles cases among HIV-infected children receiving outpatient HIV care at the BBCCCOE. Two of the children died from measles-related disease. We matched 377 control subjects who were also seen at the BBCCCOE. Median age at measles diagnosis for cases was 12.5 years (interquartile range: 9.8, 14.6 years) compared to 10.1 years (interquartile range: 7.1, 13.1 years) among controls.

In univariable analyses, cases and controls were similar with respect to WHO HIV stage, detectable HIV viral load, anti-retroviral therapy regimen and receipt of PCP prophylaxis (Table 1). Children diagnosed with measles were more likely to be older (P<0.0001), moderately or severely malnourished (P=0.03) and immunosuppressed (P=0.06). In the multivariable-adjusted model, only age remained significantly associated with measles (P <0.0001). Children in the oldest quartile (≥13.1 years) were more than four times more likely to be infected with measles compared to children in the youngest quartile (<7.1 years) (OR: 4.34; 95% CI: 2.21, 8.53).

Table 1.

Summary of descriptive, univariable and multivariable-adjusted risk factors for measles of N=566 HIV-infected children and adolescents with and without measles in Botswana, 2008-2011.

			Univariable		Multivariable
Characteristic (total n with data)	Control subjects (N=377)a	Case subjects (N=189)a	OR (95% CI)	P	OR (95% CI)	P
Male (n=566)	209 (55.4)	91 (48.2)	0.74 (0.52, 1.06)	0.10	0.87 (0.58, 1.31)	0.50
Age, in quartiles (n=566)				<0.0001		<0.0001
Q1: < 7.1 years	97 (25.7)	21 (11.1)	1 (reference)		1 (reference)
Q2: ≥7.1 to <10.3 years	89 (23.6)	32 (16.9)	1.68 (0.89, 3.18)		2.05 (1.02, 4.11)
Q3: ≥10.3 to <13.1 years	94 (24.9)	54 (28.6)	2.50 (1.43, 4.39)		3.18 (1.69, 5.97)
Q4: ≥13.1 years	97 (25.7)	82 (43.4)	3.77 (2.15, 6.60)		4.34 (2.21, 8.53)
Moderate or severe malnutritionb (n=563)	54 (14.4)	43 (22.9)	1.62 (1.06, 2.47)	0.03	1.03 (0.63, 1.68)	0.91
Receipt of PCPc prophylaxis (n=433)	69 (24.4)	37 (24.7)	0.98 (0.60, 1.59)	0.93
Antiretroviral therapy regimen (n=566)				0.67
1st line	290 (76.9)	142 (75.1)	1 (reference)
2nd,3rd, or 4th line	10 (2.7)	8 (4.2)	1.69 (0.59, 4.84)
Suspended	69 (18.3)	33 (17.5)	0.98 (0.62, 1.54)
Not yet on treatment	8 (2.1)	6 (3.2)	1.49 (0.52, 4.32)
WHO HIV stage (n=559)				0.51
1st	144 (38.8)	72 (38.3)	1 (reference)
2nd	19 (5.1)	16 (8.5)	1.53 (0.74, 3.17)
3rd	91 (24.5)	48 (25.5)	1.05 (0.65, 1.68)
4th	117 (31.5)	52 (27.7)	0.86 (0.53, 1.39)
Detectable HIV viral loadd (n=524)	39 (11.2)	21 (11.9)	0.96 (0.53, 1.73)	0.87
Degree of immune suppressione (n=541)				0.06		0.27
Not significant	297 (82.0)	137 (76.5)	1 (reference)		1 (reference)
Mild	36 (9.9)	14 (7.8)	0.81 (0.42, 1.58)		0.73 (0.35, 1.51)
Advanced	12 (3.3)	16 (8.9)	2.78 (1.23, 6.26)		1.74 (0.73, 4.17)
Severe	17 (4.7)	22 (6.7)	1.47 (0.69, 3.13)		1.63 (0.73, 3.66)

IQR, interquartile range; SD, standard deviation

^aData are n (%) unless otherwise noted.

^bModerate malnutrition defined as weight for height (<5 years of age) or BMI (≥ 5 years of age) less than two but greater than three negative standard deviations below median World Health Organization (WHO) standards; severe malnutrition defined as less than three negative standard deviations below WHO median.

^cPneumocystis pneumonia (PCP) prophylaxis (trimethoprim/sulfamethoxazole)

^dDetectable HIV viral load defined as >400 copies/mL.

^eNot significant immunosuppression defined as CD4% ≥ 35% for age 0-12 months, CD4% ≥ 25% for age 13-59 months and CD4>500 for age ≥ 5 years; mild immunosuppression defined as 35% > CD4% ≥ 25% for age 0-12 months, 25% > CD4% ≥ 20% for age 13-59 months years and 500 > CD4 ≥ 350 for age ≥ 5 years; advanced immunosuppression defined as 25% > CD4% ≥ 20% for age 0-12 months, 20% > CD4% ≥ 15% for age 13-59 months and 350 > CD4 ≥ 200 for age ≥ 5 years and severe immunosuppression defined as CD4% < 20% for age 0-12 months, CD4% < 15% for age 13-59 months and CD4 < 200 for age ≥ 5 years.

Discussion

This retrospective matched case-control study provides evidence that older, HIV-infected adolescents in Botswana are at particularly high risk of developing measles. This finding contrasts a previous study in Thailand that found no significant difference in age among HIV-infected children with and without protective measles antibody (7). The association between measles and older age may be attributable to vaccine failure. In a Zambian study, only one-half of HIV-infected children had protective levels of measles antibodies at 27 months after vaccination compared to 89% of HIV-uninfected children (5). Immune reconstitution with anti-retroviral therapy has also been shown to be insufficient in itself to restore protective levels of measles IgG (8). However, children on anti-retroviral therapy are more likely to develop protective antibodies after re-vaccination compared to children not on anti-retroviral therapy (9). We note that the majority of patients in this study were on anti-retroviral therapy during the period of exposure to measles.

Another explanation for the increased measles risk with older age could be increasing vaccination coverage over time. In 1991, when the oldest members of the study population would have received their routine measles vaccination, measles vaccination coverage in Botswana was 88% compared to 94% in 2010, when the youngest members would have been vaccinated. Unfortunately, we did not have access to individual-level measles vaccination status within the BBCCCOE electronic medical record. Future studies could examine whether the association between measles and older age is due to lack of vaccination or waning immunity.

Regardless of etiology, this study identified a high-risk group that may need additional protection against measles. Following the 2009-2010 measles epidemic in Southern Africa, many developing countries, including Botswana, incorporated a second dose of measles vaccine at 15-18 months of age into the standard childhood immunization regimen. HIV-infected older children and adolescents may benefit from alternative or additional doses following immune reconstitution with ART if measles outbreaks continue to occur.

Moderate or severe malnutrition and immunosuppression were significantly associated with measles in the univariable models (P =0.03 and P =0.06, respectively). However, in the multivariable model, the effects of both factors were attenuated and statistically not significant. Previous studies have found associations between lack of adequate measles antibodies and both malnutrition (5) and immunosuppression (10). Percentages of severely malnourished and immunosuppressed children in the current study were small.

A limitation of the study was that the diagnosis of measles was based on clinical rather than laboratory criteria. We used the WHO clinical case definition given its practical applicability to settings where measles is prevalent. Additionally, measles IgM testing was performed in a minority of individuals early in the epidemic and when the clinical picture made the diagnosis uncertain. However, we may have misclassified some severely immunosuppressed patients as non-cases as these individuals can have atypical manifestations of measles, including infection without rash. However, any misclassification of measles infection is unlikely to be differential with respect to age and therefore would bias our results towards the null.

In summary, this matched case-control study of HIV-infected children in a country with a single routine measles vaccination dose at nine months of age found that age was a significant risk factor for measles. It will be important to determine whether the introduction of a second routine dose at 15-18 months of age is sufficient to protect older HIV-infected children and adolescents during future measles outbreaks.