Mortality surveillance report in South Africa: Focus on neonatal population

The fifth Rapid Mortality Surveillance Report, 2015 by the Burden of Disease Research Unit of the South African Medical Research Council was published in December, 2016. This report provides a national picture of the trends and causes of death of the newborn population.

RMS Report 2015 Summary_image 1

This summary, written by Dr Natasha R Rhoda, Senior Neonatal Consultant at Groote Schuur Hospital in Cape Town and the chairperson of the National Perinatal Mortality and Morbidity Committee in South Africa, concentrates on data for the period 2012 to 2015 and presents the key findings of the Rapid Morality Surveillance (RMS) report 2015 in relation to perinatal and neonatal mortality. For further reading, browse the report Perinatal Deaths in South Africa, 2015, which was published in July 2017.

Acronyms and abbreviations

DHA                      Department of Home Affairs

DHIS                     District Health Information System

HDACC                 Health Data Advisory and Co-ordination Committee

HIV                        Human immunodeficiency virus

NMR                     Neonatal mortality rate

NPR                      National Population Register

PPIP                      Perinatal Problem Identification Program

RMS                      Rapid Mortality Surveillance

Stats SA               Statistics South Africa

VR                         Vital registration

Key messages

  • Rapid Mortality Surveillance (RMS) is useful for the continuous rapid assessment of the changing trends in deaths by age and sex.
  • However, the RMS system cannot reliably estimate neonatal deaths because it uses death notifications submitted to the Department of Home Affairs (DHA) and registered in the National Population Register (NPR). Only a small percentage of neonatal deaths (approximately 10% in 2011) were estimated to be included in the National Population Register, possibly because they happened before the birth was also registered.
  • Because of this, the neonatal mortality rate (NMR) up to 2015 is based on adjusted data from the District Health Information System (DHIS), rather than vital registration neonatal deaths.
  • The estimated Neonatal Mortality Rate, based on DHIS data, was 12 deaths per 1,000 live births in 2015. This means effectively that the NMR must be halved to 6 per 1,000 live births in the next two years in order to reach the 2019 target of 6 deaths per 1,000 live births.

Background

Rapid Mortality Surveillance is useful for the continuous rapid assessment of the changing trends in deaths by age and sex. Each assessment is carried out by experienced mortality researchers. The RMS system:

  • Provides information about deaths within months of the occurrence, unlike other reports with a two-year lag reporting period.
  • Tracks changes in mortality associated with the roll out of interventions, for example, the prevention of mother-to-child transmission of HIV.
  • Informs research and policy action.

Data source

In South Africa, the DHA is responsible for civil registration and the maintenance of a computerised NPR. The national vital statistics system is Statistics South Africa (Stats SA).

The source of data for the RMS systems is death notifications submitted to the DHA and registered in the NPR. The report also uses cause-of-death data from Stats SA.  It is important to note that there are differences between the numbers of deaths captured by Stats SA and the RMS system. Stats SA captures all deaths notified to the DHA, while the RMS system only captures deaths notified to the DHA which have been recorded in the NPR i.e. only the deaths of individuals with a South African birth or identity certificate (as only people with these certificates are on the Population Register). RMS, therefore, captures fewer deaths compared to Stats SA. Encouragingly the data gap between the two systems has decreased from 16% in 2000 to less than 1% in 2014 (see Table 1 and Figure 1 below). Figure 1 is available on page 3 of the report.

Table 1: Deaths recorded in the NPR / RMS in comparison to deaths recorded in STATS SA, with the data gap between the systems reported in per cent (Joubert et al 2012).

Source: NPR / RMS STATSSA STASSA>NPR % difference in total number of deaths (NPR less than STATS SA)
Year(s) Total number of deaths Difference in total number of deaths (STATS SA – NPR / RMS)
2000 359 470 416 420 56 950 16%
2001-2006 400 183 –

555 081

455 126 –

613 108

54 943 –

58 027

14% –

10%

2009 533 165 579 711 46 546 9%
2010 508 960 543 856 34 896 7%
2011 485 023 505 803 20 780 4%
2012 466 653 491 100 24 447 5%
2013 453 198 458 933 5 735 1%
2014 447 732 452 191 4 459 <1%
2015 447 244

Figure 1: NPR deaths as a proportion of Stats SA deaths by age group, 2000-2014 (Dorrington et al 2016)Figure 1_NPR 2000-14_Summary RMS Report 2015

  • Over 95% of the death notifications of people aged 25 and over (see orange line above), and around 90% of all other age groups except those in the first year of life, also known as infant deaths (see dark blue line above), are recorded in the NPR (Fig 1).
  • Only 50% of infant deaths are recorded in the NPR. This possibly refers to poor birth registration, for example, if a baby died before their birth was registered.

Too few neonatal deaths (10% in 2011) are recorded in the NPR; therefore, RMS data cannot be used to produce reliable estimates of neonatal deaths. There is also a lag in the release of the cause-of-death data. Because of this, in the RMS report, the neonatal mortality rate (NMR) up to 2015 is based on adjusted data from the District Health Information System (DHIS), rather than vital registration neonatal deaths, on the assumption that completeness of reporting of these neonatal deaths is the same as that of infant deaths (i.e. 85%).

Data trends

  • The NMR (as derived from the DHS) has declined gradually from 14 per 1 000 live births to 11 per 1 000 live births for the period 2009-2013, which was below the Health Data Advisory and Co-ordination Committee (HDACC) target. It then increased slightly to 12 per 1 000 in 2014 and 2015.
  • Impact indicator targets for mortality were set to be achieved by March 2019 as recommended by the HDACC, using 2012 values as baseline. This means effectively that the NMR must be halved to 6 per 1,000 live births in the next two years in order to reach the 2019 target.

Table 2: Key mortality indicators, RMS 2012-2015  (this information, including the footnotes, is part of a larger table on page i of the report) (Dorrington et al 2016)

MATERNAL AND CHILD MORTALITY (OUTPUT 2)
INDICATOR TARGET 20191 Baseline 20122 20132 20142 2015
Under-5 mortality rate (U5MR) per 1 000 live births 33*

(20% reduction)

41 41 40 37
Infant mortality rate (IMR) per 1 000 live births 18 27 28 28 27
Neonatal mortality rate3 (<28 6

 

11 11 3 12 3 12 3
days) per 1,000 live births
  1. Target values for 2019 were revised based on the current 2012 estimate
  2. Based on NPR data rather than VR data because of apparent significant under-recording by VR
  3. Method changed to derive directly from the DHIS neonatal deaths and birth data

* Assumed published figure of 23 is a typographical error

  • Deaths from causes originating in the perinatal period do not follow any seasonal trend unlike deaths coded without any cause which follow the pneumonia pattern over the winter months.

Overview and final comments

  • Neonatal deaths recorded in the NPR only account for a small percentage of registered deaths. The NMR, therefore, cannot be calculated using RMS data.
  • The number of neonatal deaths in the DHIS increased steadily from 2008 to2014, but it is uncertain how late neonatal deaths have contributed to this increase.
  • The number of neonatal deaths in the VR data was fairly steady from 2006 to 2009 but has declined slightly since 2009.
  • The number of neonatal deaths recorded by the DHIS has exceeded those recorded by the VR since 2012, and probably due to decline in completeness of VR.

Weaknesses and strengths of the DHIS and VR system:

  • The DHIS records all facility deaths especially in the first week of life, so the early neonatal death rate is accurate.
  • The DHIS misses deaths that occur in the private sector or in the community.
  • The VR system misses neonatal deaths that have not been registered.

Conclusion

Findings from the RMS system are regularly reported in publicly available papers, reports, and conference presentations. Detailed recording of avoidable factors for perinatal deaths is documented in the Perinatal Problem Identification Program (PPIP), which is the national stillbirth and neonatal death audit system. The National Department of Health is currently working towards combining elements of DHIS data with elements of PPIP data to streamline and strengthen the quality of data to be used to effectively plan perinatal care in South Africa.

References

Dorrington, R.E., Bradshaw, D., Laubscher, R., Nannan, N. (2016). Rapid mortality surveillance report 2015. Cape Town: South African Medical Research Council

Joubert, J., Rao, C., Bradshaw, D., Dorrington, R.E., Vos, T., Lopez, A. (2012). Characteristics, availability and uses of vital registration and other mortality data sources in post-democracy South Africa. Global Health Action, 5(1)