Summary
Background
Colombia experienced a Zika virus (ZIKV) outbreak in 2015–2016. To assist with planning for medical and supportive services for infants affected by prenatal ZIKV infection, we used a model to estimate the number of pregnant women infected with ZIKV and the number of infants with congenital microcephaly from August 2015 to August 2017.
Methods
We used nationally reported cases of symptomatic ZIKV disease among pregnant women and information from the literature on the percent of asymptomatic infections to estimate the number of pregnant women with ZIKV infection occurring August 2015–December 2016. We then estimated the number of infants with congenital microcephaly expected to occur August 2015–August 2017. To compare to the observed counts of infants with congenital microcephaly due to all causes reported through the national birth defects surveillance system, the model was time limited to produce estimates for February–November 2016.
Findings
We estimated 1140–2160 (interquartile range [IQR]) infants with congenital microcephaly in Colombia, during August 2015–August 2017, whereas 340–540 infants with congenital microcephaly would be expected in the absence of ZIKV. Based on the time limited version of the model, for February–November 2016, we estimated 650–1410 infants with congenital microcephaly in Colombia. The 95% uncertainty interval for the latter estimate encompasses the 476 infants with congenital microcephaly reported during that approximate time frame based on national birth defects surveillance.
Interpretation
Based on modeled estimates, ZIKV infection during pregnancy in Colombia could lead to 3–4 times as many infants with congenital microcephaly in 2015–2017 as would have been expected in the absence of the ZIKV outbreak.
Funding
This publication was made possible through support provided by the Bureau for Global Health, U.S. Agency for International Development under the terms of an Interagency Agreement with Centers for Disease Control and Prevention.
Keywords
Background
Zika virus (ZIKV) is a flavivirus in the same family as dengue, yellow fever, and West Nile viruses. It was first isolated from a nonhuman primate in 1947 in Uganda.
1
Since the first recognized outbreak in the Yap Islands in 2007,2
ZIKV has spread rapidly through seronaïve populations. The ZIKV outbreak in continental South America was first detected in Brazil in May 2015.3
Recognizing the severity of the outbreak and the potential implications in Brazil, the Instituto Nacional de Salud (INS) of Colombia began official surveillance of ZIKV in August 2015. The INS first reported autochthonous transmission in October 2015 when a cluster of nine laboratory confirmed ZIKV cases were identified in Northern Colombia. ZIKV disease has since been detected throughout Colombia, with a peak of reported suspected cases at the national level in pregnant women occurring in late January 2016 (epidemiologic week 4).4
In April 2016, the Centers for Disease Control and Prevention (CDC) concluded that prenatal ZIKV infection is a cause of congenital microcephaly and absent or poorly developed brain structures.
5
, 6
There are no known interventions for preventing microcephaly or other fetal brain anomalies following ZIKV infection during pregnancy. The first confirmed ZIKV-linked infant with congenital microcephaly in Colombia was reported in April 2016,7
and while an increase in the number of infants with congenital microcephaly has been observed,Instituto Nacional de Salud. El Boletín Epidemiológico Semanal 2016 Epidemiologic Week 13. http://www.ins.gov.co/boletin-epidemiologico/Boletn%20Epidemiolgico/2016%20Boletín%20epidemiológico%20semana%2013.pdf.
8
the number of birth defects associated with ZIKV infection during pregnancy is expected to continue to increase because of the delay between infection during pregnancy and the observation of ZIKV-associated birth defects.9
As of December 31, 2016, Colombia had recorded 106,659 suspected cases of ZIKV disease in Colombia, of which 19,716 were among pregnant women.
10
Most reported ZIKV cases, including those among pregnant women, were not laboratory confirmed; 6363 (32.3%) pregnant women with reported ZIKV disease had confirmed ZIKV infection by real-time reverse transcription polymerase chain reaction (rRT-PCR) laboratory testing. Recorded cases of microcephaly in Colombia are determined to be attributable to maternal ZIKV infection based on confirmatory laboratory testing and the absence of other known causes.Instituto Nacional de Salud. El Boletín Epidemiológico Semanal 2016 Epidemiologic Week 52. http://www.ins.gov.co/boletin-epidemiologico/Boletn%20Epidemiolgico/2016%20Boletín%20epidemiológico%20semana%2052.pdf.
4
There may be cases of microcephaly unaccounted for in Colombia's surveillance system.To assist with planning for appropriate medical and supportive services for affected families, we used reported cases of suspected ZIKV disease among pregnant women, combined with information on the percentage of infections that are estimated to be asymptomatic, to estimate the number of pregnant women with ZIKV infection since the start of the ZIKV outbreak in Colombia through December 31, 2016. We then estimated the number of infants born August 2015–August 2017 with congenital microcephaly as a result of ZIKV infection among pregnant women. The model was time limited to produce estimated microcephaly case counts among pregnancies ending February–November, 2016. We then compared the modeled estimates to the number of infants with congenital microcephaly reported via the Colombian national birth defects surveillance system during that time period.
8
Methods
We developed a model to estimate the number of pregnant women in Colombia with ZIKV infections and the number of infants affected with microcephaly that would be expected to occur from the initiation of official ZIKV disease surveillance in August 2015–August 2017. We summarize the model input parameters in Table 1.
Table 1Model input parameters for estimating total number of infants with congenital microcephaly expected from maternal Zika virus (ZIKV) infection during pregnancy in Colombia.
| Parameter | Definition | Estimate | Range | Uncertainty distribution | |
|---|---|---|---|---|---|
| Full model | Time limited model | ||||
| Incidence of ZIKV infection in pregnant women | Expected overall percent of pregnant women infected with ZIKV as of December 31, 2016 2 , 10 ,Instituto Nacional de Salud. El Boletín Epidemiológico Semanal 2016 Epidemiologic Week 52. http://www.ins.gov.co/boletin-epidemiologico/Boletn%20Epidemiolgico/2016%20Boletín%20epidemiológico%20semana%2052.pdf. 18 , 30
| High | 21–27% | 17–22% | Uniform |
| Medium | 12–16% | 10–13% | |||
| Low | 6–8% | 5–7% | |||
| Very low | 1–2% | 1–2% | |||
| Birth rate | 2014 birth rate 12
Departamento Administrativo Nacional de Estadística. Births by area and sex. http://www.dane.gov.co/index.php/estadisticas-por-tema/salud/nacimientos-y-defunciones/nacimientos/nacimientos-2014. | High: 14.5 per 1000 | N/A | N/A | |
| Medium: 16.3 per 1000 | |||||
| Low: 13.0 per 1000 | |||||
| Very low: 13.2 per 1000 | |||||
| Population | 2015 population 11
Departamento Administrativo Nacional de Estadística. Population Projections. 2015. http://www.dane.gov.co/reloj. | High: 12,573,511 | N/A | N/A | |
| Medium: 5,584,629 | |||||
| Low: 7,639,987 | |||||
| Very low: 22,4705,278 | |||||
| ZIKV-associated risk of congenital microcephaly 18 , 19
| Infection during the first trimester 18
| 6.50% | 1–13% | Uniform | |
| Infection during the second trimester 18
| 0.03% | 0–0.2% | Triangular | ||
| Infection during the third trimester 18
| 0.00% | 0–0.2% | Triangular | ||
| Baseline risk of congenital microcephaly | Risk of congenital microcephaly in the absence of ZIKV virus infection during pregnancy 12
Departamento Administrativo Nacional de Estadística. Births by area and sex. http://www.dane.gov.co/index.php/estadisticas-por-tema/salud/nacimientos-y-defunciones/nacimientos/nacimientos-2014. | 2/10,000 live births | 2–6/10,000 live births | Triangular | |
a Estimates vary by strata of risk of ZIKV infection in pregnant women and estimates are derived from observed Sivigila ZIKV disease surveillance data.
b Strata information is detailed in Table 2.
c In the absence of ZIKV.
d Full model estimated counts of infected pregnant women and pregnancy outcomes for pregnancies completed April 2016–August 2017; Time limited model estimated counts of infected pregnant women and pregnancy outcomes for pregnancies completed February to November 2016.
We assumed that all women living in one of Colombia's 32 departments or Bogotá, Distrito Capital, who were pregnant at any point during the course of the ZIKV outbreak, August 2015–December 2016, were at risk for ZIKV infection and, given infection, at risk for having an infant born with microcephaly. We estimated the departments’ birth rates based on the population sizes and annual number of births in 2014.
11
, Departamento Administrativo Nacional de Estadística. Population Projections. 2015. http://www.dane.gov.co/reloj.
12
We assumed that contraceptive usage did not change during the outbreak, and, as implied by our use of population sizes and birth rates from 2014, that the number of pregnancies did not decrease.Departamento Administrativo Nacional de Estadística. Births by area and sex. http://www.dane.gov.co/index.php/estadisticas-por-tema/salud/nacimientos-y-defunciones/nacimientos/nacimientos-2014.
We estimated the cumulative probability of maternal ZIKV infection based on the cumulative incidence of reported ZIKV disease among pregnant women as of December 31, 2016.
10
By utilizing cumulative incidence, any effects of seasonality and the natural progression of the ZIKV epidemic are averaged over the time period of our analysis. We obtained cumulative incidence from weekly clinical incidence reports from Colombia's national public health surveillance system (Sistema de Vigilancia en Salud Pública, or Sivigila) summarizing new cases of ZIKV disease among pregnant women by department. The incident ZIKV disease cases reported to Sivigila were comprised of those that reported symptoms, with or without laboratory confirmation, to a health care center.Instituto Nacional de Salud. El Boletín Epidemiológico Semanal 2016 Epidemiologic Week 52. http://www.ins.gov.co/boletin-epidemiologico/Boletn%20Epidemiolgico/2016%20Boletín%20epidemiológico%20semana%2052.pdf.
4
During the time period considered in our analysis, Colombia was primarily utilizing rRT-PCR testing for ZIKV RNA for confirmatory testing only for those cases in high-risk groups, including pregnant women with testing within 5 days of symptom onset.Estimating the risk of ZIKV infection by administrative departments
Roughly half of Colombia's population lives in high elevation areas, where the climate is cooler, and arbovirus transmission is unlikely to occur.
13
Therefore, we used the administrative department, the first-level subdivision of the country, as the unit of analysis since the reported incidence of ZIKV disease in pregnant women varied widely across the country.10
However, within each department, we assumed a uniform risk of maternal ZIKV infection.Instituto Nacional de Salud. El Boletín Epidemiológico Semanal 2016 Epidemiologic Week 52. http://www.ins.gov.co/boletin-epidemiologico/Boletn%20Epidemiolgico/2016%20Boletín%20epidemiológico%20semana%2052.pdf.
14
We also categorized departments into four strata by their reported cumulative incidence of ZIKV disease among pregnant women: >5% (high), 3–5% (medium), 1–2% (low) and <1% (very low) (Table 2). The combined total population size was approximately 12.6 million in the high-risk stratum, 5.6 million in the medium-risk stratum, 7.6 million in the low-risk stratum and 22.5 million in the very low-risk stratum.Table 2Categorization of departments and districts in Colombia, by reported incidence of Zika virus (ZIKV) disease in pregnant women August 2015–December 2016.
| ZIKV disease strata | Births per 1000 population 12
Departamento Administrativo Nacional de Estadística. Births by area and sex. http://www.dane.gov.co/index.php/estadisticas-por-tema/salud/nacimientos-y-defunciones/nacimientos/nacimientos-2014. | Reported ZIKV disease – pregnant women | % of pregnant women with symptoms potentially due to ZIKV disease | Estimated % of pregnant women with symptomatic ZIKV infection | Estimated % of pregnant women with ZIKV infection | |
|---|---|---|---|---|---|---|
| High | Norte de Santander | 15.0 | 2920 | 14•3 | 8.0–10.5 | 43.3–56.5 |
| Arauca | 15.3 | 320 | 8.0 | 4.5–5.8 | 24.2–31.5 | |
| Casanare | 18.1 | 489 | 7.6 | 4.2–5.5 | 22.9–29.8 | |
| Huila | 17.9 | 1423 | 6.9 | 3.9–5.0 | 20.8–27.1 | |
| Meta | 17.1 | 1076 | 6.5 | 3.7–4.8 | 19.8–25.8 | |
| Valle de Cauca | 12.0 | 3255 | 5.9 | 3.3–4.3 | 17.8–23.1 | |
| Tolima | 11.9 | 977 | 5.8 | 3.3–4.3 | 17.6–23.0 | |
| Atlántico | 17.2 | 2137 | 5.1 | 2.8–3.7 | 15.3–19.9 | |
| TOTAL | 14.5 | 12,597 | 6.9 | 3.9–5.0 | 21–27 | |
| Medium | Caquetá | 15.9 | 378 | 4.9 | 2.7–3.6 | 14.8–19.3 |
| Santander | 15.4 | 1373 | 4.3 | 2.4–3.2 | 13.1–17.0 | |
| Archipiélago de San Andrés y Providencia | 11.1 | 34 | 4.0 | 2.2–2.9 | 12.1–15.8 | |
| Magdalena | 18.1 | 866 | 3.8 | 2.1–2.8 | 11.5–15.0 | |
| Córdoba | 16.4 | 1034 | 3.7 | 2.1–2.7 | 11.1–14.5 | |
| TOTAL | 16.3 | 3685 | 4.0 | 2.3–3.0 | 12–16 | |
| Low | Sucre | 18.2 | 448 | 2.9 | 1.6–2.1 | 8.7–11.4 |
| Putumayo | 11.6 | 112 | 2.8 | 1.6–2.0 | 8.5–11.0 | |
| Amazonas | 20.3 | 42 | 2.7 | 1.5–2.0 | 8.2–10.7 | |
| Cesar | 20.9 | 541 | 2.5 | 1.4–1.8 | 7.6–9.9 | |
| Risaralda | 12.1 | 195 | 1.7 | 0.9–1.2 | 5.1–6.7 | |
| Vichada | 11.9 | 13 | 1.5 | 0.8–1.1 | 4.6–6.0 | |
| La Guajira | 15.3 | 207 | 1.4 | 0.8–1.0 | 4.3–5.6 | |
| Cundinamarca | 8.3 | 307 | 1.4 | 0.8–1.0 | 4.2–5.4 | |
| Quindío | 10.7 | 80 | 1.3 | 0.7–1.0 | 4.0–5.2 | |
| Guaviare | 10.2 | 13 | 1.1 | 0.6–0.8 | 3.5–4.5 | |
| TOTAL | 13.0 | 1958 | 2.0 | 1.1–1.4 | 6–8 | |
| Very low | Vaupés | 9.3 | 4 | 0.9 | 0.6–0.7 | 3.0–3.9 |
| Antioquia | 11.8 | 566 | 0.7 | 0.4–0.5 | 2.3–2.9 | |
| Caldas | 10.5 | 75 | 0.7 | 0.4–0.5 | 2.2–2.9 | |
| Cauca | 11.3 | 101 | 0.6 | 0.4–0.5 | 2.0–2.5 | |
| Bolívar | 16.6 | 212 | 0.6 | 0.3–0.4 | 1.8–2.4 | |
| Bogotá, D.C. | 15.0 | 442 | 0.4 | 0.2–0.3 | 1.1–1.5 | |
| Boyacá | 12.5 | 51 | 0.3 | 0.2–0.2 | 1.0–1.3 | |
| Chocó | 12.7 | 11 | 0.2 | 0.1–0.1 | 0.5–0.7 | |
| Guainía | 14.2 | 1 | 0.2 | 0.1–0.1 | 0.5–0.7 | |
| Nariño | 10.7 | 13 | 0.1 | 0.0–0.1 | 0.2–0.3 | |
| TOTAL | 13.2 | 1476 | 0.5 | 0.3–0.4 | 1–2 |
a Cases, clinically suspected or laboratory confirmed, reported to Sivigila, as of epidemiological week 52, 2016.
b Cases, clinically suspected or laboratory confirmed, of pregnant women reported to Sivigila divided by the estimated total number of pregnant women by department/district, as of epidemiological week 52, 2016.
c Adjusted for symptomatic cases due to other infections.
d Adjusted for asymptomatic cases by (% pregnant women with ZIKV disease) * 1:4.4.
We assumed that all pregnancies resulting in a live born infant occurring in the model estimation window were uniformly distributed across months. This assumption implies that one third of the women who are pregnant in any given month would be in each of the first, second, or third trimester. Using these assumptions, we estimated the number of women in each trimester of pregnancy during the outbreak for each ZIKV disease stratum and, based on those estimates, the number of women infected with ZIKV in each trimester of pregnancy.
Adjusting for non-ZIKV infections
The symptoms of ZIKV disease are nonspecific and are similar to other mosquito borne infections, i.e., dengue and Chikungunya. In Colombia, testing by rRT-PCR of a subset of the total population, including pregnant women, found 73% of tested individuals with ZIKV symptoms were positive for ZIKV.
4
In a subgroup comprised of pregnant women with symptoms consistent with ZIKV disease (n = 582), 56% tested positive for ZIKV by rRT-PCR.3
To reflect this lack of specificity in ZIKV-related symptoms as an indicator of true infection, we assumed that between 56% and 73% of the number of pregnant women reported to Sivigila as having ZIKV symptoms were actually infected with the virus.Adjusting for asymptomatic ZIKV infections
Individuals with symptomatic ZIKV disease are reported to Sivigila, but many ZIKV infections are unreported because symptoms are absent or mild. Mothers with asymptomatic ZIKV infection have given birth to infants with microcephaly;
4
, 15
therefore, we assumed that the risk of microcephaly given maternal ZIKV infection is the same for symptomatic and asymptomatic women.6
, 16
,17
To account for underreporting attributable to asymptomatic maternal infections, we used data from previous ZIKV outbreaks. A seroprevalence study in the Yap Islands reported a mean estimated ratio of symptomatic to asymptomatic ZIKV infection of 1:4.4.2
Consequently, in our model, we estimated the incidence of ZIKV infection by adjusting the estimated percentage of pregnant women with symptoms due to ZIKV by a factor of 5.4 to account for asymptomatic infections.Estimating incidence of ZIKV infection among pregnant women
The estimates of ZIKV infection incidence among pregnant women used as parameters for the model (Table 1) were derived by first estimating the proportion of pregnant women captured in Sivigila who reported ZIKV symptoms within each ZIKV incidence stratum based on their department of residence. This proportion was then multiplied by the assumed adjustment, described above, relating presence of ZIKV symptoms due to other infections (56–73%). The resulting estimated range of true ZIKV infection among symptomatic pregnant women was then multiplied by 5.4 to adjust for the assumed ratio of total ZIKV infection in the population to that among those reporting symptoms (Table 2). Details on the approach used to derive these estimates are given in the Technical Appendix.
Model assumptions
Within each risk stratum, we summarized plausible values for the ZIKV cumulative incidence as a uniform distribution using the calculated ranges (Table 1). We assumed that susceptibility to ZIKV infection was no different for pregnant women than for the general population. We further assumed that a woman infected with ZIKV in a given month would develop immunity to ZIKV and not be susceptible to infection in subsequent months. Additionally, we assumed that ratio of symptomatic to asymptomatic cases was the same for pregnant women in Colombia as for the general population of the Yap Islands. The risk of ZIKV-associated microcephaly given maternal infection was estimated to be 1–13% for women infected in their first trimester based on observations from the ZIKV outbreaks in French Polynesia and Bahia, Brazil.
18
, 19
To reflect our belief that the true value for this risk is equally likely to fall anywhere between 1–13%, we used a uniform uncertainty distribution. Previous analyses have estimated the risk of microcephaly given maternal infection to be 0.03% (range: 0–0.2%) in the second and 0% (range: 0–0.2%) in the third trimester;18
, 19
we used a triangular distribution of uncertainty to concentrate probability closer to the assumed most likely values. Estimates of infants with congenital microcephaly not associated with ZIKV infection were based on the baseline risk of microcephaly in Colombia in the absence of ZIKV, 2 to 6 cases per 10,000 live births.12
Departamento Administrativo Nacional de Estadística. Births by area and sex. http://www.dane.gov.co/index.php/estadisticas-por-tema/salud/nacimientos-y-defunciones/nacimientos/nacimientos-2014.
Monte Carlo simulations were used to propagate the assumed uncertainty in each parameter through to the estimated outcomes. In this process, each parameter was repeatedly sampled from the assumed uncertainty distribution and outcomes were calculated for each sampled set of parameters. The model outputs were summarized using the median, the interquartile range (IQR) which represents the middle 50% of the 95% uncertainty interval, and the 95% uncertainty interval (UI), of 100,000 Monte Carlo simulations. The IQRs are presented as the primary model results as high uncertainty in model parameters leads to widely dispersed outcomes and the IQRs represent the most likely outcomes.
All statistical analyses were conducted in SAS version 9.3. Details on the model, assumptions, and the Monte Carlo process are available in the Technical Appendix.
Limiting the model to outcomes occurring February–November 2016
INS and CDC published observed case counts of infants and fetuses with microcephaly for pregnancies ending January 31–November 12, 2016.
8
To compare model estimates to the observed number of infants with congenital microcephaly, we limited the model to correspond to the time window. First, we limited the modeled pregnancies to those that would be delivered full term prior to November 2016. Next, we limited the cumulative ZIKV infection rate to infections occurring prior to the end of May 2016 (epidemiologic week 21).20
This time point was selected as it corresponds to the last month in which a woman in her 1st trimester of pregnancy could deliver a full term infant by the end of November 2016. For consistency, departments retained their strata classification from the model using the reported cumulative incidence of ZIKV disease among pregnant women. Additional details on the model adjustments used to compare estimated and observed case counts are described in the Technical Appendix.Instituto Nacional de Salud. El Boletín Epidemiológico Semanal 2016 Epidemiologic Week 21. Instituto Nacional de Salud http://www.ins.gov.co/boletin-epidemiologico/Boletn%20Epidemiolgico/2016%20Boletín%20epidemiológico%20semana%2021.pdf.
Results
According to Sivigila reports, the cumulative incidence of symptomatic ZIKV disease as of December 31, 2016 was 3.0% among pregnant women. The number and percent of pregnant women with reported ZIKV disease in the high-risk stratum was 12,597 (6.9%); in the medium-risk stratum, 3685 (4.0%); in the low-risk stratum, 1958 (2.0%); and in the very low-risk stratum, 1476 (0.5%).
10
Instituto Nacional de Salud. El Boletín Epidemiológico Semanal 2016 Epidemiologic Week 52. http://www.ins.gov.co/boletin-epidemiologico/Boletn%20Epidemiolgico/2016%20Boletín%20epidemiológico%20semana%2052.pdf.
After adjustment for symptoms due to non-ZIKV infections and accounting for asymptomatic maternal ZIKV infections, the estimated percent of pregnant women infected with ZIKV in the high-risk stratum was 21–27%, in the medium-risk stratum 12–16%, in the low-risk stratum 6–8%, and in the very low-risk stratum 1–2% (Table 2).
We estimated a total of 340–540 (IQR infants with congenital microcephaly (95% UI: 260–730) would occur nationwide in the absence of the ZIKV outbreak from August 2015 to August 2017. For each risk stratum, we estimated the median number or infants with congenital microcephaly among women with ZIKV infection during pregnancy: 80 (IQR: 40–110; 95% UI: 10–170) in the very-low risk stratum; 120 (IQR: 70–180; 95% UI: 20–240) in the low-risk stratum; 230 (IQR: 130–320; 95% UI: 40–430) in the medium-risk stratum; and 780 (IQR: 450–1110; 95% UI: 150–1460) in the high-risk stratum. Combining across strata, we estimated that the country as a whole might expect a total of 1200 (IQR: 700–1720; 95% UI: 240–2220) infants born with microcephaly to women with ZIKV infection since the initiation of surveillance through August 2017 (Table 3). In total, combining baseline risk and risk among women infected with ZIKV during pregnancy, we estimated that Colombia could expect 1650 (IQR: 1140–2160; 95% UI 630–2720) infants with congenital microcephaly since the initiation of surveillance through August 2017.
Table 3Estimated number of infants born with congenital microcephaly by Zika virus (ZIKV) disease strata, since the initiation of official surveillance through August 2017.
| ZIKV disease strata | Reported % of pregnant women with ZIKV disease | Estimated % of pregnant women with ZIKV infection | Estimated number of infants born with congenital microcephaly in absence of ZIKV infection IQR | Estimated number of infants born with congenital microcephaly associated with maternal ZIKV infection | Estimated total number of infants born with congenital microcephaly | ||||
|---|---|---|---|---|---|---|---|---|---|
| Median | IQR | 95% UI | Median | IQR | 95% UI | ||||
| High | >5 | 21–27 | 90–150 | 780 | 450–1100 | 150–1460 | 900 | 570–1230 | 270–1590 |
| Medium | 3–5 | 12–16 | 50–70 | 230 | 130–320 | 40–430 | 290 | 190–380 | 100–500 |
| Low | 1–2 | 6–8 | 50–80 | 120 | 70–180 | 70–180 | 190 | 140–240 | 80–310 |
| Very low | <1 | 1–2 | 150–240 | 80 | 40–110 | 10–170 | 270 | 220–330 | 150–430 |
| Country wide | – | – | 340–540 | 1200 | 700–1720 | 240–2220 | 1650 | 1140–2160 | 630–2720 |
a Cumulative incidence of ZIKV disease cases reported among pregnant women.
b Range of cumulative incidence of ZIKV infection cases among pregnant women, adjusted for asymptomatic cases.
c Interquartile range.
Results of limiting the model to outcomes occurring February–November 2016
When limiting the model to February–November 2016 in order to compare to available Colombian case counts for infants and fetuses with microcephaly, we estimated the total number of infants with congenital microcephaly among women with ZIKV infection during pregnancy: 120 (IQR: 90–160; 95% UI: 60–210) in the very low-risk stratum; 110 (IQR: 70–150; 95% UI: 30–210) in the low-risk stratum; 180 (IQR: 110–250; 95% UI: 50–340) in the medium-risk stratum; and 600 (IQR: 360–840; 95% UI: 140–1100) in the high-risk stratum. Combining across strata, we estimated a total of 1030 (IQR: 650–1410; 95% UI: 300–1790) infants with microcephaly among pregnant women with ZIKV infection during February–November 2016 (Table 4). For the period of January 31–November 12, 2016 there were 476 reported infants and fetuses with congenital microcephaly in Sivigila: 120 in the very low-risk stratum; 84 in the low-risk stratum; 80 in the medium-risk stratum; and 192 in the high-risk stratum. The reported number of infants and fetuses with congenital microcephaly were within the IQR of modeled estimates in the very low and low strata and were within the 95% uncertainty interval for the medium stratum, the high stratum, and country-wide.
Table 4Comparison of estimated and reported number of infants born with congenital microcephaly by Zika virus (ZIKV) disease strata.
| ZIKV disease strata | Estimated % of pregnant women with ZIKV infection | Estimated total number of infants with congenital microcephaly, February–November 2016 | Reported number of infants and fetuses with congenital microcephaly, January 31–November 12, 2016 8 | ||
|---|---|---|---|---|---|
| Median | IQR | 95% UI | |||
| High | 17–22 | 600 | 360–840 | 140–1100 | 192 |
| Medium | 10–13 | 180 | 110–250 | 50–340 | 80 |
| Low | 5–7 | 110 | 70–150 | 30–210 | 84 |
| Very low | 1–2 | 120 | 90–160 | 60–210 | 120 |
| Country wide | – | 1030 | 650–1410 | 300–1790 | 476 |
a Cumulative incidence of ZIKV disease cases reported among pregnant women from the full model.
b Range of cumulative incidence of ZIKV infection cases among pregnant women, adjusted for asymptomatic cases and case ascertainment window.
c Interquartile range.
Discussion
Our modeled estimates indicate that Colombia is likely to experience a substantial increase in the number of infants born with congenital microcephaly, with approximately 1650 (IQR: 1140–2160) infants to be born since the initiation of surveillance through August 2017. This reflects an increase of 3 to 4 times over the baseline microcephaly counts in Colombia prior to the ZIKV outbreak based on a comparison of IQRs, which accounts for the uncertainty distribution surrounding the median values. This increase may pose a significant challenge to local authorities to ensure that medical and support services are available for infants born with microcephaly and their families.
The modeled estimates developed to reflect the time window matching that of observed microcephaly case counts tended to have medians of the uncertainty distributions greater than the number reported by Colombia for the period from January 31 to November 12, 2016 (Table 4). However, at both the country-level and all risk strata levels, the 95% UI of the estimated values included the observed case counts. A limitation of this comparison is that the observed infants with congenital microcephaly were among infants and fetuses whereas our model estimates are only for live born infants. Our model estimates would have been even higher had fetal losses with microcephaly been included. Despite the uncertainties reflected in the modeling approach, this suggests that the estimates produced for the full time frame, since the initiation of surveillance through August 2017, using the model are likely to be higher than, but still not inconsistent with, the number of infants with congenital microcephaly that may be reported by the surveillance system in that time frame. There are a number of explanations for why the model estimates exceed those based on birth defects surveillance. First, birth defects surveillance will miss ZIKV infected mothers and infants tested after the acute infection has resolved, when evidence of infection becomes undetectable by rRT-PCR laboratory tests, which may lead to a misclassification as non-ZIKV associated congenital microcephaly.
8
Second, it is probable that some infants with microcephaly were not reported to Sivigila, which utilizes passive birth defects surveillance methods.8
Third, it is possible that some pregnancy losses with microcephaly were not diagnosed or not reported to national surveillance. Fourth, the model assumes an unchanged pregnancy rate over time, and there are data to suggest that Colombia's pregnancy rate went down following the 2015–2016 ZIKV outbreak.8
Uncertainty remains regarding the incidence of maternal ZIKV infection in Colombia. Even within strata, reported incidence varied substantially across departments, presumably influenced both by differences in ZIKV transmission rates by area and differences in the completeness of reporting. The denominator for our cumulative infection estimates does not exactly match the numerator because the number of ZIKV infection cases is among all pregnant women while the denominator is the estimated total number of live births; not all pregnancies result in live birth. As noted, there is substantial uncertainty regarding most of the parameters in our model. We accommodated this uncertainty by using ranges of plausible values and a Monte Carlo process to sample from assumed uncertainty distributions. Therefore, the estimates have wide uncertainty intervals that should be interpreted carefully, with the extremes representing possible, but unlikely values.
In constructing the model, we made several assumptions that influenced our estimation of the number of infants we expect will be born with microcephaly. First, while slight variations in the number of births per month occur in Colombia, we assumed that births were uniformly distributed throughout the year. We were therefore able to estimate the outcomes based on estimated overall maternal infections since the initiation of surveillance and for the outcomes occurring during February–November 2016. In Colombia, reported symptomatic ZIKV incidence has been decreasing since late January 2016, indicating that the risk of infection has declined. Our use of cumulative incidence averages the risk of ZIKV infection across the time period of analysis. Consequently it is unnecessary to account in our model for the seasonality of ZIKV infection or the monthly progression of the ZIKV epidemic through the population. Our estimation of the number of pregnant women with ZIKV infection was partly founded upon data based on rRT-PCR testing of symptomatic individuals,
3
which misses infections in persons tested after symptoms resolved or were asymptomatic. Therefore, our model may have underestimated the number of pregnant women infected with ZIKV.Microcephaly has been recognized as a severe outcome of congenital ZIKV infection and the spectrum of other adverse outcomes includes brain abnormalities with or without microcephaly, neural tube defects and other early brain malformations, eye abnormalities without mention of a brain abnormality, and other consequences of central nervous system dysfunction.
5
,15
, 16
, 17
,21
, 22
, 23
, 24
, 25
, 26
, 27
, 28
, 29
This model is only estimating the number of infants with microcephaly at birth, thus the total number of infants with birth defects associated with maternal ZIKV infection during pregnancy will likely be greater than these estimates. Though the modeled estimates may be higher than observed case counts, they are a useful complement to surveillance data for understanding service needs, in that they approximate the upper bound for the range of the number of infants with congenital microcephaly born during the analysis period. While information is limited and key uncertainties persist in the model, ZIKV infection during pregnancy in Colombia could be responsible for 3 to 4 times as many infants with congenital microcephaly born since the initiation of surveillance through August 2017 compared to the number that would have been expected had the ZIKV outbreak not occurred. These modeled estimates may assist public health agencies and healthcare providers with planning for appropriate medical and supportive services for affected families.Research in context
Evidence before this study
Several studies on the impact of Zika virus (ZIKV) infection have been published presenting evidence of an increase of birth defects in infants with maternal ZIKV infection during pregnancy. In April 2016, the Centers for Disease Control and Prevention (CDC) established that prenatal ZIKV infection is a cause of congenital microcephaly and absent or poorly developed brain structures. Colombia experienced a ZIKV outbreak in 2015–2016 which resulted in nearly 20,000 cases of Zika virus disease reported among pregnant women.
Added value of this study
The Instituto Nacional de Salud (INS) of Colombia provides observed counts of pregnant women with symptomatic Zika virus disease and infants with congenital microcephaly. The INS utilizes a passive surveillance systems for reporting congenital microcephaly and some cases of microcephaly may be unreported. Our model estimates the number of infants with congenital microcephaly by using the number of pregnant women infected with ZIKV and the risk of congenital microcephaly associated with maternal ZIKV infection. The modeled estimates suggest that ZIKV infection during pregnancy in Colombia could lead to 3 to 4 times as many infants with congenital microcephaly in 2015–2017 as would have been expected in the absence of the ZIKV outbreak.
Implications of all the available evidence
Comparing our modeled estimates to reported data on the number of cases of microcephaly in Colombia between February and November of 2016, our estimates were substantially higher than the reported counts, though the reported counts did fall within the 95% uncertainty interval of the modeled estimate. These modeled estimates may assist public health agencies and healthcare providers with planning for appropriate medical and supportive services for affected families. Models, such as the one presented here, can be used to estimate impacts of future outbreaks.
CDC disclaimer
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention, the U.S. Agency for International Development, nor the Instituto Nacional de Salud.
Appendix. Supplementary materials
References
- Epidemiological notes on some viruses isolated in Uganda; Yellow fever, Rift Valley fever, Bwamba fever, West Nile, Mengo, Semliki forest, Bunyamwera, Ntaya, Uganda S and Zika viruses.Trans R Soc Trop Med Hyg. 1953; 47: 13-48
- Zika virus outbreak on Yap Island, federated states of micronesia.N Engl J Med. 2009; 360: 2536-2543
- Cumulative Zika suspected and confirmed cases reported by countries and territories in the Americas.(Accessed February 24, 2016)
- Zika virus disease in colombia - preliminary report.N Engl J Med. 2016;
- Zika virus and birth defects–reviewing the evidence for causality.N Engl J Med. 2016; 374: 1981-1987
- Birth defects among fetuses and infants of US women with evidence of possible zika virus infection during pregnancy.JAMA. 2017; 317: 59-68
Instituto Nacional de Salud. El Boletín Epidemiológico Semanal 2016 Epidemiologic Week 13. http://www.ins.gov.co/boletin-epidemiologico/Boletn%20Epidemiolgico/2016%20Boletín%20epidemiológico%20semana%2013.pdf.
- Preliminary report of microcephaly potentially associated with Zika virus infection during pregnancy - colombia, January-November 2016.MMWR Morb Mortal Wkly Rep. 2016; 65: 1409-1413
- Projecting month of birth for at-risk infants after Zika virus disease outbreaks.Emerg Infect Dis. 2016; 22: 828-832
Instituto Nacional de Salud. El Boletín Epidemiológico Semanal 2016 Epidemiologic Week 52. http://www.ins.gov.co/boletin-epidemiologico/Boletn%20Epidemiolgico/2016%20Boletín%20epidemiológico%20semana%2052.pdf.
Departamento Administrativo Nacional de Estadística. Population Projections. 2015. http://www.dane.gov.co/reloj.
Departamento Administrativo Nacional de Estadística. Births by area and sex. http://www.dane.gov.co/index.php/estadisticas-por-tema/salud/nacimientos-y-defunciones/nacimientos/nacimientos-2014.
- Revision to CDC's Zika travel notices: minimal likelihood for mosquito-borne Zika virus transmission at elevations above 2000 meters.MMWR Morb Mortal Wkly Rep. 2016; 65: 267-268
- Presencia de Aedes (Stegomyia) aegypti (Linnaeus, 1762) y su infección natural con el virus del dengue en alturas no registradas para Colombia.Biomédica. 2016; 36: 303-308
- Zika virus infection and stillbirths: a case of hydrops fetalis, hydranencephaly and fetal demise.PLoS Negl Trop Dis. 2016; 10e0004517
- Vital signs: update on Zika virus-associated birth defects and evaluation of all U.S. infants with congenital Zika virus exposure - U.S. Zika pregnancy registry, 2016.MMWR Morb Mortal Wkly Rep. 2017; 66: 366-373
- Pregnancy outcomes after maternal Zika virus infection during pregnancy - U.S. territories, January 1, 2016-April 25, 2017.MMWR Morb Mortal Wkly Rep. 2017; 66: 615-621
- Association between Zika virus and microcephaly in French Polynesia, 2013-15: a retrospective study.Lancet. 2016; 387: 2125-2132
- Zika and the risk of microcephaly.N Engl J Med. 2016; 375: 1-4
Instituto Nacional de Salud. El Boletín Epidemiológico Semanal 2016 Epidemiologic Week 21. Instituto Nacional de Salud http://www.ins.gov.co/boletin-epidemiologico/Boletn%20Epidemiolgico/2016%20Boletín%20epidemiológico%20semana%2021.pdf.
- Baseline prevalence of birth defects associated with congenital Zika virus infection - massachusetts, north carolina, and atlanta, georgia, 2013-2014.MMWR Morb Mortal Wkly Rep. 2017; 66: 219-222
- Zika Virus infection in pregnant women in rio de janeiro.N Engl J Med. 2016; 375: 2321-2334
- Detection and sequencing of Zika virus from amniotic fluid of fetuses with microcephaly in Brazil: a case study.Lancet Infectious Dis. 2016; 16: 653-660
- Zika virus infection with prolonged maternal viremia and fetal brain abnormalities.N Engl J Med. 2016; 374: 2142-2151
- Computed tomographic findings in microcephaly associated with Zika virus.N Engl J Med. 2016; 374: 2193-2195
- Zika virus associated with microcephaly.N Engl J Med. 2016; 374: 951-958
- Zika virus intrauterine infection causes fetal brain abnormality and microcephaly: tip of the iceberg?.Ultrasound Obstet Gynecol. 2016; 47: 6-7
- Ocular findings in infants with microcephaly associated with presumed Zika virus congenital infection in salvador, brazil.JAMA Ophthalmol. 2016; 134: 529-535
- Evidence of perinatal transmission of Zika virus, French Polynesia, December 2013 and February 2014.Euro Surveill. 2014; 19
- Transmission dynamics of Zika virus in island populations: a modelling analysis of the 2013-14 french polynesia outbreak.PLoS Negl Trop Dis. 2016; 10e0004726
Article info
Publication history
Published online: April 05, 2018
Accepted:
February 25,
2018
Received in revised form:
February 20,
2018
Received:
September 21,
2017
Identification
Copyright
Published by Elsevier Ltd on behalf of The British Infection Association.
