Study sites and sample

The data for this study were obtained from the VASA survey, which was undertaken by the National Population Commission (NPC) from February to December 2014. This was the fourth country that conducted a VASA study using technical assistance from the Maternal and Child Epidemiology Estimation (MCEE) (formerly known as the Child Health Epidemiology Reference Group or CHERG), and this study built upon the experiences had by Malawi, Niger, and Cameroon when conducting smaller VASAs. The study protocol and other methodological considerations have been previously described in detail [6].

In Nigeria, the VASA interviews were conducted with women in the 38,522 households in which deaths were identified using 2013 Nigerian Demographic and Health Survey (NDHS) data [2]. All deaths that occurred during the prior 10 years among the children of surveyed women aged 15–49 years were identified via the NDHS. To limit issues related to faulty recall while obtaining an adequate sample size, the VASA study examined deaths among children aged up to 59 months within a 5-year recall period. A total of 1,206 neonatal (0–27 days old) deaths and 2,779 child (1–59 months old) deaths were identified from 2009 to 2013. However, to minimize the burden placed on respondents and because deceased children within the same household were likely to be similar in social and healthcare system factors, the VASA interview focused on one neonatal or child death selected at random within each participating household. This strategy allowed for the inclusion of 986 neonatal and 2,268 child deaths (for a total of 3,254 deaths). Based on an expected proportion of 0.50, these samples sizes enabled confidence intervals of ±0.03 and ±0.02 to be established around the point estimates for the most common causes of neonatal and young child mortality, respectively. Of the 3,254 households sampled, 2,944 (90.5%) had completed the VASA interview; of these cases of mortality, 164 were stillbirths, 723 were neonatal (0–27 days) deaths, and 2,057 were child (1–59 months) deaths. Of the 310 uncompleted interviews, 185 (59.7%) were in households in the northern regions that were not visited because of safety concerns. Restricted working hours, not receiving clearance to enter the clusters on a regular basis, and potential threats were some of the challenges faced by the study teams. The current analysis included only the 2,057 child deaths.

Data collection tools and the VASA interview

The VASA questionnaire is a combination of the Population Health Metrics Research Consortium (PHMRC) verbal autopsy questionnaire (focused on the biomedical cause of death) and the CHERG Pathway Analysis social autopsy (SA) questionnaire [7] (focused on the well-child and illness events leading up to a death). The CHERG SA questionnaire is a version of an earlier Pathway Analysis instrument [8] that has been updated to more fully elucidate the household, community, and healthcare system factors associated with neonatal and child mortality, including the maternal, preventive, and curative care provided in the event of a neonatal death or stillbirth.

The questionnaire was translated into the following three major languages spoken by most persons in the study area: Yoruba, Igbo, and Hausa. Three local anthropologists were recruited to perform the translations, and another team of experienced staff at the NPC independently back-translated and compared the translations to the original English questionnaire. Discrepancies were then scrutinized to determine the source of the errors and corrected via consultations between the anthropologists and the back translators. An English version of the VASA questionnaire is available in S1 File.

Finally, the translations were inserted into the CSPro software application that was developed to enable direct, field-based Computer Aided Personal Interview (CAPI) capture of the VASA interview data on a netbook computer.

The technical team members who were trained during the pretest/training of trainers (TOT) phase of the project received English interview training and training in the three languages. The trainings were conducted following best practice training procedures, including classroom lectures and presentations, daily reviews, mock interviews, class exercises, and a written test at the end of every module. The trainings also included conducting practice field interviews with mothers who had suffered a recent neonatal or child death. Remedial classes were established for those who did not do well on the tests. A special training was conducted for the supervisors, which focused on the work assignments for team members and backup and transfer of completed cases from the field to the IT manager at the NPC headquarters.

Fourteen interviewing teams carried out the VASA fieldwork, with 11 teams placed in the northern and 3 teams placed in the southern region. The difference in the number of teams per region was due to the greater number of cases identified in the North. The teams carried out field exercises in all 36 states of the country and the Federal Capital Territory (FCT). Each team consisted of a supervisor, three to four interviewers (who were predominantly women), and one driver. The fieldwork was initiated November 1, 2014 and continued through December 2014.

Data quality was ensured by technical team/trainers who also functioned as quality controllers. Data quality was also monitored via the field check tables that were generated by the CSPro program on the supervisors’ netbooks as fieldwork progressed. This methodology was advantageous because the technical team/trainers (or field coordinators) could advise field teams of problems detected during fieldwork. Specifically, tables were generated to evaluate various data quality parameters. The technical team/trainers met in Abuja at the end of each monitoring visit to discuss fieldwork issues and then travelled to states where immediate attention was required. Representatives from USAID and NPC’s commissioners and state directors also monitored the fieldwork process.

Data analysis

The current analysis of preventive and curative care data followed procedures described previously [9] and was guided by the coverage of the following key indicators along the continuum of the preventive child care provided both inside and outside the home: (a) sociodemographic and household factors associated with child mortality, and (b) illness recognition and care-seeking patterns encompassed by the Pathway to Survival model [7, 8, 10]. All examined interventions have been shown to be efficacious and effective in promoting child survival and, thus, included among the interventions examined by the Lives Saved (LiST) tool [10] or recommended by the World Health Organization (WHO); therefore, these interventions should be accessible to all children. A list and the definitions of some of the operational variables used throughout this paper are available in the article written by Koffi et al. [11].

In addition, a scoring system was developed based on the reported feeding behaviors, activity levels and mental status of the children to assess the impact of perceived illness severity at onset on caregivers’ attempts to seek care for their children. Hence, the following three independent categories of illness severity were constructed: normal/mild, moderate, and severe. Each of these categories had a range of possible scores. The details of the method have been described in a prior paper [11], and the definitions are provided in S1 Fig.

The 2013 NDHS, which served as the platform for the VASA study, used a multistage, stratified sampling methodology with strata defined based on state, rurality and cluster. Therefore, we used design-based weighting to ensure the sample was representative of the overall population of Nigerian women. We also used the Taylor linearization method for variance estimation.

Statistical analyses were conducted using the survey commands (svy) in STATA 14.0 (StataCorp, College Station, TX) to adjust for the cluster sampling design, weighting, and calculation of standard errors. For regional analyses, we followed the country’s standard geopolitical divisions, which group the 36 states and the FCT into the following six zones: Northwest, Northeast, North-Central, Southeast, Southwest and South-South. These zones comprised the following two regions: the northern region (herein, the North, i.e., Northwest, Northeast, North-central) and the southern region (herein, the South, i.e., Southeast, Southwest and South-South).

Ethical considerations

In Nigeria, the VASA study was first approved by the National Health Research Ethics Committee of the Federal Ministry of Health (FMOH) and then approved by the Johns Hopkins School of Public Health’s Institutional Review Board. All respondents provided informed consent before the interview was conducted.

Demographic and household characteristics

Table 1 shows the sociodemographic characteristics of the deceased children overall and in the northern and southern regions. Significant associations were identified between several of the sociodemographic variables and childhood mortality and region (all p<0.001 in chi square tests). The median age at illness onset was 11 months in the South and 12 months in the North (median of 12 months overall). More specifically, the majority of the deaths in the South (46.5%) occurred during the post-neonatal period (1–11 months of age), with 22.4% and 24.1% of deaths occurring in the 1–5 month and 6–11 month age groups, respectively. In the North, the proportion of deaths increased incrementally from 14.1% in the 1–5 month age group to 42.7% in the 24–59 month age group. The duration of fatal illness was shorter in the southern region (median = 5 days) than in the northern region (median = 7 days). The gender distributions of the deceased children were similar between the two regions, with 51% boys and 49% girls identified in both the North and the South. Many of the deceased children in the South (66.9%) were born at a healthcare facility, while only 19.1% of the children born in the North had a facility-based birth. More than three quarters (77.6%) of children in the North and more than half (54.2%) of children in the South died at home.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Characteristics of 2057 deceased children.

https://doi.org/10.1371/journal.pone.0177025.t001

Table 2 shows that almost all the evaluated characteristics of the mother, her domestic partner, and the household were significantly associated with region (p<0.001). More than two-thirds (71.5%) of the mothers of deceased children married early in adolescence (<16 years) in the North, whereas only 5.9% of mothers married early in the South. Many the mothers in both regions were 25 years or older when their index child died (57.2% in the North and 76.9% in the South). Women who had no formal schooling constituted a higher percentage (72.5%) of the mothers in the North than in the South (11.1%). The majority (47.2%) of the mothers in the South had some secondary education. The data revealed a similar pattern in the association between paternal education and region. The majority of the deceased children in the northern region resided in the poorest households in the country. Conversely, 80.5% of the deceased children in the South lived in middle-income and richer households. More than three-fourths (76.4%) of the deceased children lived in households located in rural areas. The urban/rural distribution varies greatly by region; the majority of households (86.3%) in which a deceased child resided in the North were located in the rural areas, while 40.0% of deceased children in the South resided in a rural area. Furthermore, of the 1,395 households in which deceased children resided in rural areas of the north, 83.6% were in the two bottom quintiles of wealth, whereas 23.5% of the 176 rural households in which deceased children resided in the South were in the bottom two wealth quintiles. Among households that experienced a child death, the median household sizes were 7 and 5 members in the North and South, respectively. The median time required to travel from the caregivers’ household to their usual healthcare center was 30.0 minutes in both the North and South.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Characteristics of the mother and her household, 2057 child deaths.

https://doi.org/10.1371/journal.pone.0177025.t002

Preventive home care

Fig 1 displays the various types of preventive home care received by the deceased children prior to his/her fatal illness. Approximately three in four children (73.8%) were not likely to be exposed to smoke (i.e., he/she was usually away from the mother when she cooked inside the house), with a significantly higher proportion of children reported to avoid smoke exposure in the South than in the North (79.8% versus 72.2%, χ2 (1) = 10.4, p = 0.033). Less than one third of the children always slept under an insecticide-treated bed net before their fatal illness began, and no significant regional difference was identified. Overall, among children in whom illness was first identified at 0–23 months (n = 1,299), 46% were being fed appropriately at the time their fatal illness began, and no difference was observed between the North (46.0%) and the South (44.0%). In both the northern and the southern regions, the proportions of children whose illness symptoms first appeared at 0–23 months and were exclusively breastfed decreased incrementally by age group at illness onset (0–5 months: North 26.0%, South 33.8%; 6–11 months: North 11.4%, South 7.0%; 12–23 months: North 4.7%, South 0.1%). Fig 1 shows the percentage of deceased children aged 12–59 months (n = 1,359) who had been vaccinated against each of the six major preventable childhood diseases by the age of one year. Overall, only 8.0% of the children were fully immunized against these diseases before they reached their first birthday, and a significant difference was observed between the immunization rate in the northern (6.0%) and southern (22.0%) region (χ2 (1) = 67.1, p<0.001). The polio vaccine had the highest rates of coverage in the North (ranging from 53% to 59% for one to three doses), and in the South, BCG, DPT or PENTA, and measles demonstrated the highest rates of coverage, ranging from 51% to 87%. Deceased children in the North were unlikely to be fully immunized against DPT or PENTA by the age of one year (only 11% had received all three doses).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Coverage along the continuum of care for young children in Nigeria, from 2009–2013.

Notes: *: Proportion of children who were NOT usually beside or carried by their mother when she cooked inside the home. **: Insecticide-treated bed net. ***: Children whose fatal illness started at 0–23 months and satisfied either of the following conditions: (i) Child’s illness began before 6 months of age (0–5 months), he/she was being breastfed at the time of fatal illness and was not given anything but breast; (ii) Breastfed children whose fatal illness started at 6–8 months old and 9–23 months old who received, respectively, at least two and three complementary non-liquid feedings each day; (iii) Child's fatal illness started at 6–23 months old and he/she received at least four replacement feeds each day (including milk and solid, semi-solid and soft foods) milk as food. ****Information on immunizations was obtained either from the vaccination card or when there was no written record, from the respondent (mainly the mother). Polio0 is the Polio vaccination given at birth; Fully Immunized children received BCG, measles, and three doses each of DPT/PENTA and polio vaccine (excluding polio vaccine given at birth).

https://doi.org/10.1371/journal.pone.0177025.g001

Curative care seeking

Fig 2 exhibits the steps and possible breakdowns in the Pathway to Survival that may have contributed to mortality in the included children. Table 3 depicts the distribution of these indicators by region and the results of chi-square tests for their corresponding associations. These analyses included only the children whose caretakers provided information on the types of curative actions taken for the illness (i.e., n = 1,615 in the North and n = 441 in the South). When the caregivers of these 2,056 children first noticed that the child was ill, nearly all (97.1%) of them recognized the presence of severe or possibly severe signs or symptoms. Healthcare was obtained or sought outside the home for almost all (84.6%, n = 1,739) the children; 63.9% of the caregivers of deceased children in the North first sought or attempted to seek outside care, while only 51.5% did so in the South (p<0.001). A significantly greater proportion of caregivers in the South (38.1% vs. only 19.4% in the North; p<0.001) first sought healthcare inside the home and later sought or attempted to seek formal healthcare. In total, more than half (53.9%) of the children in the South and 44.5% of children in the North received or had caregivers who attempted to seek formal care only; additionally, another 20.4% and 15.6% of children received or had caregivers who attempted to seek both informal and formal care in the South and the North, respectively. In other words, 74.3% of children in the South and 60.1% of children in the North received or had caregivers who attempted to seek formal care (χ2 (1) = 20.3, p<0.001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Pathway to survival component and indicators.

https://doi.org/10.1371/journal.pone.0177025.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The “Pathway to Survival” component and indicators.

https://doi.org/10.1371/journal.pone.0177025.g002

In addition, more than three-quarters (75.5%, 68/90) of both formal and informal healthcare professionals who provided care to the children or from whom caregivers attempted to seek care in the South consisted of pharmacists or drug sellers, while these healthcare professionals only comprised 44.9% of providers reported in the North (113/252), (χ2 (1) = 25.3, p<0.001). Conversely, 45.5% (170/374) and 34.3% (23/67) of caregivers sought only informal care from pharmacists or drug sellers in the North and the South, respectively. Overall, when informal care was sought for the children’s illnesses, more than half (58.0%, 91/157) of caregivers in the South and 45.2% (283/626) of caregivers in the North (χ2 (1) = 7.8, p = 0.016) sought care from pharmacists or drugs sellers.

The majority of the children were already moderately to severely ill at illness onset (Fig 3); 81.1% of the children who did not receive any care were moderately to severely ill, and more than three-quarters of children whose caregivers provided only home care were moderately to severely ill at illness onset. Furthermore, 23.4% (21/91) of children who did not receive any care were severely ill, while 11.1% (207/1878) of children who received any care had an illness of similar severity (χ2 (1) = 12.9, p<0.001).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Perceived illness severity at illness onset by care-seeking patterns, Nigeria 2009–2013.

https://doi.org/10.1371/journal.pone.0177025.g003

The median duration (delay) from illness onset until seeking formal care was 2 days among children whose caregivers sought or attempted to seek both informal and formal care and children whose caregivers sought or attempted to seek formal care only, and no regional difference was observed (χ2 (1) = 0.0, p = 0.905). Children who first received home care (but whose caregivers later sought or attempted to seek some formal care) experienced a longer delay prior to seeking formal care (median = 3 days, IQR = 2–6 days]) than did children whose caregivers first sought or attempted to seek formal care (median = 2 days, IQR = 1–3 days) (χ2 (1) = 144.8, p<0.001).

However, among those who sought formal care, the decision to do so occurred a median of 25.4 hours in the North and 24.0 hours in the South after illness onset (Fig 4). A median 2-day delay in seeking formal care was observed regardless of the perceived severity of illness (χ2 (2) = 4.1, p = 0.131).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Illness severity ranking at onset and at decision to seek care among children for whom caregivers tried to seek or sought some formal care (N = 1293*), by region, Nigeria 2009–2013.

Note: *5 children had missing information that did not allow their illness severity ranking.

https://doi.org/10.1371/journal.pone.0177025.g004

Furthermore, among children whose caregivers sought or attempted to seek formal care (n = 970 and n = 328, excluding the 4 and 1 children with missing perceived illness severity data in the north and south, respectively), the proportion perceived to be severely ill increased from 11.9% at the time of illness onset to 19.2% when the decision was made to seek formal care in the North and from 7.6% to 15.0% in the South (Fig 4).

Although nearly three-quarters (74.3%) of caregivers in the South (versus 60.1% in the north) reported that they attempted to seek some formal care on behalf of their sick children, they were less likely to reach the healthcare facility than were those in the north (χ2 (1) = 88.6, p<0.001); reasons for this decreased likelihood included child death before departure, child death en route, and inability to reach the healthcare provider. The proportion of children who were moderately ill at the time when their caregiver decided to seek care on their behalf and died before reaching their healthcare provider was greater in the South (77.4%; albeit not significantly, χ2 (1) = 4.8, p = 0.079) than in the North (66.6%). On the other hand, the proportion of severely ill children who died before reaching their healthcare provider was higher in the North (24.5%) than in the South (14.7%) but not significantly so (χ2 (1) = 5.0, p = 0.060).

Overall, 901 (n = 727 in the North; n = 174 in the South) of the 1,298 children whose caregivers attempted to seek formal care reached a healthcare provider before the child died. The healthcare providers reached first were, in descending order of frequency, hospitals (n = 474), non-governmental (NGO) or governmental (GOV) clinics (n = 255), private clinics (n = 90), and trained community health providers (n = 83) (Fig 5). The distributions of the following first reached healthcare providers differed by region: hospitals (49.7% in the North, 64.4% in the South), NGO or GOV clinics (32.9% in the North, 9.1% in the South), private clinics (9.3% in the North, 13.0% in the South), and trained community health providers (8.1% in the North, 13.5% in the South). The proportion of children who first reached hospitals was significantly higher in the South than in the North (χ2 (1) = 12.5, p = 0.004), but the proportion first reaching NGO or GOV clinics was significantly greater in the North than in the South (χ2 (1) = 40.2, p<0.001). The delay in obtaining formal care seeking did not vary significantly by the type of healthcare provider reached first (χ2(3) = 3.2, p = 0.368).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Type of first formal health provider reached by caregivers with their ill children.

https://doi.org/10.1371/journal.pone.0177025.g005

Nearly all (94.7%) the children were moderately or severely ill when they reached their first healthcare provider. No statistically significant regional difference was identified in perceived illness severity among this subset of children.

More than half (58.9%) of the children who reached a private clinic and almost half (47.0%) of those who reached a hospital died at those health facilities (Fig 5). Only one in four (26.1%) children who reached a private clinic and 18.0% of children who reached a hospital were perceived to be severely ill at the time their caregivers decided to seek care on their behalf.

The proportion of children who reached their first healthcare provider’s facility and died there and were severely sick when their caregivers decided to seek formal care was significantly higher in the North (30.2%) than in the South (18.3%) (χ2 (1) = 5.0, p = 0.046). However, the overall proportion of children who reached their first healthcare provider and died there was significantly higher in the South (21.5%) than in the North (14.5%) (χ2 (1) = 12.6, p = 0.008) (Table 3).

More children in the North (30.6%) than in the South (17.9%) left their first healthcare provider alive (χ2 (1) = 27.8, p<0.001). Of the 574 children who left their first provider alive, 95.6% (n = 548; or n = 477 in the North and n = 71 in the South) were moderately or severely sick at discharge. Only 11.9% (n = 68) of the children were referred to another healthcare provider (some also received home care recommendations), with the following statistically significant difference observed between regions: 10.5% (50/477) in the North and 23.2% in the South (17/71), (χ2 (1) = 9.6, p = 0.007). According to the caregivers, the majority of the children (59 out of 68, or 86.8%) were referred to a second provider because their first provider was not capable of managing the problem, while 24.5%-26.5% believed their child’s condition required supplies (e.g., drugs, IV, oxygen) and/or equipment (e.g., X-ray machine) that were not available at the first facility during the child’s visit, and 3 (4.6%) did not know the reason their child was referred to a second provided (results not shown). Fifty-six (56/68 = 82.5%) sought care from the referred second health provider. More than 40% of the children received only home care recommendations from their first healthcare provider at discharge.

Fig 6 focuses on the 548 children who had moderate or severe illness at discharge from the first healthcare provider. Trained health workers were most likely to refer these children to another healthcare provides (18.1%), while private clinic facilities were the least likely (3.9%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Decision of first healthcare provider for children moderately or severely sick at discharge (N = 548), by health facility/provider.

https://doi.org/10.1371/journal.pone.0177025.g006

Furthermore, 70% (n = 384) of the moderately/severely ill children the not receive additional formal healthcare either because their caregiver did not attempt to reach this provided at all (n = 342) or they tried but did not reach the provider before they died (n = 42). Fourteen percent (n = 77) reached another health provider and left the facility alive, while 16% (n = 87) reached another provided and died at their facility. we estimated how long after leaving the first provider the 342 and 77 children died, and the former died 2 days (median) (IQR: 1–5 days) after leaving the first healthcare provider, and the latter died a median of 8 days after first discharge (median) (IQR: 4–14 days) (χ2 (1) = 86.9, p<0.001).

Similarly, of the 87 children who died at the second health provider, 29 were not admitted to that facility; hence, we assumed they died immediately after reaching the facility. We also added the number of days the other children (n = 58) stayed at the latter health facility to the time elapsed between seeking care from the first and second providers and approximated that these children died a median duration of 4 days (median) (IQR = 2–11 days) after leaving the first provider. The results of the chi-square test indicated that the two medians (4 days versus 2 days) were significantly different (χ2 (1) = 22.7, p<0.001).

Barriers to seeking healthcare

Fig 7 shows the barriers to seeking care for fatal illnesses for all children except those who died immediately (n = 87). Caregivers who did not take or attempt to take their children to a formal healthcare provider (n = 671) were asked to describe what prevented them from going to the healthcare provider, while those who did take their child to receive healthcare (n = 1,298) were asked what constraints they had to overcome during this process. In the North, the proportions of caregivers reporting barriers were similar among those whose children did and did not receive care. However, in the South, caregivers who did not take their children to visit a healthcare provider reported these problems more frequently (18%-47%) than those who did (1%-31%). Overall, cost (34.2%), lack of transportation (24.3%), and distance (15.1%) were the most commonly cited constraints to seeking care from a healthcare provider. Cost was an especially common complaint among those in the South who did not seek care for their children (47%). Approximately 15% of respondents whose children were not taken to receive care stated that they did not think that the child was sick enough to seek formal healthcare. This practice was especially common in the South (25.2%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Main care-seeking constraints for child deaths (N = 1969).

https://doi.org/10.1371/journal.pone.0177025.g007