Regional variation in sudden unexpected death in infancy in New Zealand

Abstract

Aim

To estimate the relative risk of sudden unexpected death in infancy (SUDI) by district health board (DHB) in New Zealand after adjustment for socio‐economic deprivation, ethnicity and other demographic factors.

Methods

We conducted a population‐based cohort study using data from the Integrated Data Infrastructure, a large research database containing linked data from a range of government agencies. The study population was all live births and their mothers in New Zealand from 2012 to 2018. The exposure of interest was DHB. The outcome was SUDI.

Results

There were 418 068 live births in New Zealand from 2012 to 2018, and of these 415 401 (99.4%) had valid DHB data. There was considerable variation in the proportion of infants in each DHB living in the most deprived decile varying from 4.5% in Nelson, West Coast and Canterbury to 29.7% in Counties Manukau. There were 267 SUDI cases, giving an overall rate of 0.64/1000 live births during the study period (2012–2018). The SUDI rate varied from 1.11/1000 in Northland to 0.30/1000 in Waitemata and Auckland. Counties Manukau had the largest number of deaths (n = 54; rate = 1.08/1000). Five DHB regions had increased risk of SUDI compared to the reference group but, after adjustment, no DHB was significantly increased.

Conclusions

This study found that there is marked variation in SUDI risk by DHB, but this is explained by socio‐economic and demographic variation within DHBs. This study emphasises the importance of the contribution of social determinants of health to SUDI.

What is already known on this topic

1. Geographic variation in sudden infant death syndrome (SIDS) mortality has been reported in many countries.

2. In New Zealand, there was a north‐to‐south gradient with highest levels of SIDS in the south of the South Island and lowest levels in the north of the North Island, but this gradient is no longer seen.

3. High rates of sudden unexpected death in infancy (SUDI) occur in district health boards (DHBs) with greater socio‐economic disadvantage and Māori and Pacific people.

What this paper adds

1. This study found that there is marked variation in SUDI risk by DHB, but this is explained by socio‐economic and demographic variation within DHBs.

2. This study emphasises the important contribution of social determinants of health in SUDI.

In 1969, the term sudden infant death syndrome (SIDS) was suggested and defined as ‘The sudden death of any infant or young child, which is unexpected by history, and in which a thorough post‐mortem examination fails to demonstrate an adequate cause for death’. ¹ The recognition of prone sleeping position and bed sharing were important risk factors for SIDS led to some of these deaths being certified as positional asphyxia and accidental suffocation or strangulation in bed. At the same time, some pathologists argued that SIDS was a diagnosis by exclusion and was not a cause of death preferring to use the terms ill‐defined or cause unknown. ² All these certifiable causes of death are captured using the rubric sudden unexpected death in infancy (SUDI; the term SUID is used in the United States), which closely equates with the use of SIDS in the 1970s to 1990s. In effect, the term SIDS in the 1970s to 1990s and the current use of SUDI are interchangeable.

Geographic variation in post‐neonatal and SIDS mortality has been reported in many countries, including New Zealand. In New Zealand in the 1980s, there was reported a north–south gradient, with low rates in the north of the North Island gradually increasing to the highest rates in the south of the South Island. ³ , ⁴ The SIDS mortality rates in health districts correlated with mean average temperatures for those health boards. ⁴ We found that prone sleeping position was associated with a higher risk of SIDS in the South Island (OR = 7.0) compared with the North Island (OR = 3.0). ⁵ This and the observation that prone sleeping position varied with season, excess bedding and clothing, markers of illness and altitude suggested that prone sleeping position was related to SIDS through a thermal mechanism. ⁵ With the marked reduction in prone sleeping position following the recommendation that babies should be placed supine to sleep (‘Back to Sleep’ campaign), the north–south gradient disappeared, but has been replaced with higher rates of mortality in district health boards (DHB) that have a higher proportion of Māori and Pacific people and socio‐economic deprivation. In the years 2015–2019, Counties Manukau DHB was reported to have the highest number of SUDI cases in the post‐neonatal age group (n = 42) more than twice the number of the next DHB (Canterbury, n = 19). ⁶ The SUDI mortality rate was significantly higher in Counties Manukau DHB (1.00/1000 live births) than the national average (0.66/1000), whereas Canterbury DHB was lower (0.59/1000) illustrating the danger of comparisons of absolute numbers only.

The Child and Youth Mortality Review Committee (CYMRC) in New Zealand is a statutory committee which is accountable to the Health Quality & Safety Commission. It advises the Commission on how to reduce preventable deaths of New Zealand children and young people aged 28 days to 24 years. CYMRC examined SUDI mortality in detail from 2002 to 2015. ⁷ This showed a significant decline in SUDI, especially over the last 4 years examined. Although SUDI rates have declined in all ethnic groups, the SUDI rate in Māori has consistently been higher than non‐Māori, non‐Pacific (predominantly European). ⁶

The CYMRC report stated that ‘Many factors influence SUDI rates in DHBs, including service delivery and the characteristics of the population the DHB serves’. ⁷ Certainly, socio‐economic deprivation and ethnicity appear to be major factors influencing SUDI mortality, but the influence of differences in service delivery has not been reported.

As Counties Manukau DHB has the highest number of SUDI cases, it is probably not surprising that attention has focussed on this DHB. Radio NZ made an Official Information Act request for information about 19 SUDI deaths that occurred in Counties Manukau DHB in 2019 and for any briefing sent to the Minister of Health regarding these deaths. ⁸ Furthermore, there has been direct criticism of Counties Manukau DHB ‘While [Counties Manukau DHB] has ample funding to prevent SUDI, there hasn't been a prioritisation for Māori in a culturally appropriate way’. ⁹

The aim of this study was to examine variation in SUDI mortality by DHB in New Zealand. We hypothesised that the higher rates in some DHBs are explained by population differences in socio‐economic deprivation, ethnicity, and other demographic factors.

Methods

The study design is a population‐based cohort study of all births and their mothers occurring in New Zealand from 2012 to 2018. These years were chosen as SUDI mortality had decreased and had plateaued over these years. ¹⁰ Reporting of this study follows the STrengthening the Reporting of OBservational studies in Epidemiology Statement (STROBE). ¹¹

The cohort

The cohort data were sourced from the Integrated Data Infrastructure (IDI), a large research database containing linked data about people and households, maintained by Stats NZ. ¹² , ¹³ , ¹⁴ IDI is population‐based administrative data, covering the ever‐resident population of NZ, so virtually the entire population of New Zealand is covered. It would be anticipated that for infants born in New Zealand, the coverage would be very close to 100%. Data are from a range of NZ government agencies, including Statistics, the Ministry of Health, and the Ministry of Social Development. Data have been linked at the individual level, ¹⁴ which allows for a cohort of children and their families to be tracked over time. False positive errors are estimated to be <2% for all agencies. ¹⁵ Data on all births were linked with maternity data and other datasets within the IDI.

Definition and origin of exposure variables of interest

The exposure variable of interest is DHB:

The DHB where the infant resides was obtained from the DHB of domicile from baby records in the maternity collection. There were 20 DHBs and as some were small, contiguous DHBs were amalgamated as follows: (i) Northland, (ii) Waitemata and Auckland, (iii) Counties Manukau, (iv) Waikato, (v) Bay of Plenty, Tairāwhiti, Lakes and Hawkes Bay, (vi) Taranaki, Whanganui, MidCentral and Wairarapa, (vii) Capital & Coast and Hutt, (viii) Nelson, West Coast and Canterbury, and (ix) South Canterbury and Southern.

Outcome: sudden unexpected death in infancy

Mortality data from the Ministry of Health were used to identify SUDI cases, which were infants who died under 1 year of age and whose cause of death was certified as due to any one of the following International Classification of Diseases, 10th Revision (ICD10) codes ¹⁶ :

R95 (SIDS),

R96 (other sudden death, cause unknown),

R99 (other ill‐defined and unspecified cause of mortality),

W75 (accidental suffocation and strangulation in bed),

W78 (inhalation of gastric contents), and

W79 (inhalation and ingestion of food, causing obstruction of respiratory track) were categorised as SUDI.

Confounders and covariates

The analysis adjusted for the following variables, which are known to be associated with SUDI ¹⁷ , ¹⁸ and were in the linked IDI datasets: birthweight, gestational age, ethnicity of the child, maternal marital status, maternal age, parity, sex of infant, maternal smoking status, socio‐economic deprivation index, receipt of any social support benefit and maternal education. The definition, origin and categorisation of these variables:

Birthweight (g): Birthweight was obtained from birth records and was categorised as <1500, 1500–2499, 2500–2999, 3000–3499, 3500–3999 and 4000+.

Gestational age (completed weeks): Gestational age was obtained from birth records and was categorised <28, 28–31, 32–36, 37, 38, 39, 40 and 41+.

Prioritised ethnicity of the child: Parental‐reported ethnicity was obtained from the birth records. Using the Ministry of Health's ethnicity data protocols, children were assigned into an ethnic group using the following hierarchy of prioritisation: (i) Māori, (ii) Pacific, (iii) Asian, (iv) Middle Eastern, Latin American, and African (MELAA) and (v) European and other. ¹⁹

Marital status: The maternal characteristics were collected at registration for antenatal care. The mother's marital status was obtained from the birth records, and was categorised as married, de facto, single or not stated.

Maternal age: Maternal age was obtained from the birth records and is her age at the time of birth of the infant. It was categorised <20, 20–24, 25–29, 30–34 and 35+ years.

Number of previous births: Categorised as 0, 1, 2 and 3+.

Sex of infant: This was obtained from the birth records.

Smoking status: The primary source was the maternity data which contains information on smoking at 2 weeks after delivery. Data from the 2013 Census were used to fill in missing data. Never and ex‐smoker were categorised as non‐smoker, and regular as smoker. Missing smoking data from both the maternity data and Census were categorised as missing. 95.2% of smoking data came from the maternity data, 2.9% from 2013 Census and 1.9% were missing.

Socio‐economic deprivation index (NZDep): Socio‐economic deprivation was estimated using the New Zealand Index of Deprivation 2013, based on the deprivation characteristics of ‘meshblocks’ (small areas with a typical population of 60–110 people). ²⁰ The New Zealand Index of Deprivation 2013 combines 2013 census data relating to income, home ownership, employment, qualifications, family structure, housing, access to transport and communications into a single measure of relative socio‐economic deprivation. Each meshblock is assigned a score from 1 (least deprived) to 10 (most deprived), with the same number of meshblocks in each of the 10 categories. These deciles were then categorised as deciles 1–4 = 1 (least deprived), deciles 5 and 6 = 2, deciles 7 and 8 = 3, decile 9 = 4 and decile 10 = 5 (most deprived). The score is assigned according to maternal place of residence.

Benefit: Receipt of any of the following benefits, using data from the Ministry of Social Development Benefit Dynamics dataset: Unemployment Benefit, Unemployment Benefit Hardship, Unemployment Benefit Training, Job Seeker, Emergency Benefit, Emergency Maintenance Allowance, Invalids Benefit, Sickness Benefit, Widows Benefit, Sole Parent Support, Youth/Young Parent Payment, Domestic Purposes Benefit‐Woman Alone, Domestic Purposes Benefit‐Caring for Sick or Infirm and Supported Living Payment.

Education level: Categorised as No qualification, School qualification, Post‐school qualification below degree, University degree or higher.

Statistical analysis

Statistical analyses were performed using SAS Enterprise Guide version 8.3 (SAS Institute Inc., Cary, NC, USA). Adjusted relative risks (aRRs) and their 95% confidence intervals (CIs) were calculated using PROC GENMOD with Poisson distribution and log link with SUDI as the outcome and DHB as the exposure variable. Multivariable models controlled for all the potential confounding variables described above. All counts were randomly rounded to base 3 as per the confidentiality requirements of Statistics New Zealand. ²¹

Results

The cohort data included all live births occurring in New Zealand from 2012 to 2018, derived from Department of Internal Affairs data (n = 418 068). Of these, 415 401 (99.4%) had valid information on DHB.

The characteristics of the cohort are shown in the Table 1. The characteristics of the exposure variable (DHB) are shown in Table 2. As expected, there was considerable variation. The proportion living in the most deprived decile varied from 4.5% in Nelson, West Coast and Canterbury to 29.6% in Counties Manukau. The proportion of the births that were Māori varied from 15.8% in Waitemata and Auckland to 59.4% in Northland, and proportion of Pacific births varied from 2.7% in Northland to 29.2% in Counties Manukau. The proportion of mothers who smoked varied from 6.2% in Waitemata and Auckland to 27.2% in Northland.

Table 1.

Characteristics of the cohort (all live births in New Zealand, 2012–2016). Counts have been random‐rounded to be divisible by 3 as per the confidentiality requirements of Stats NZ

	Total cohort
	n	%
District Health Board
Northland	15 210	3.7%
Waitemata and Auckland	96 537	23.3%
Counties Manukau	58 008	14.0%
Waikato	37 440	9.0%
Bay of Plenty, Tairāwhiti, Lakes and Hawke's Bay	50 559	12.2%
Taranaki, Whanganui, MidCentral and Wairarapa	34 470	8.3%
Capital & Coast and Hutt	38 460	9.3%
Nelson Marlborough, West Coast and Canterbury	56 076	13.5%
South Canterbury and Southern	28 374	6.8%
Neighbourhood deprivation (higher = greater deprivation)
Deciles 1–4	135 546	32.4%
Deciles 5 and 6	79 224	19.0%
Deciles 7 and 8	87 879	21.0%
Decile 9	50 523	12.1%
Decile 10	61 038	14.6%
Unknown	3585	0.9%
Ethnicity
Asian	67 839	16.2%
European and Other	180 864	43.3%
Middle Eastern, Latin American, African	7179	1.7%
Maori	119 055	28.5%
Pacific	42 861	10.3%
Marital status
Married	219 867	52.6%
De facto	115 428	27.6%
Single	70 668	16.9%
Unknown	11 832	2.8%
Maternal age (y)
<20	20 064	4.8%
20–24	70 557	16.9%
25–29	113 037	27.1%
30–34	126 567	30.3%
35+	87 432	20.9%
Maternal smoking
Non‐smoker	351 453	84.1%
Smoker	58 518	14.0%
Unknown	7824	1.9%
Maternal education
No qualification	31 956	7.6%
School qualification only	111 162	26.6%
Post‐school qualification below degree	72 144	17.3%
Tertiary degree	100 767	24.1%
Unknown	101 766	24.4%
Benefit recipient
No	206 490	49.5%
Yes	211 038	50.5%
Number of previous births
0	213 015	51.0%
1	128 136	30.7%
2	49 599	11.9%
3+	27 042	6.5%
Sex of infant
Female	203 400	48.7%
Male	214 353	51.3%
Birthweight (g)
<2500	24 771	5.9%
2500–2999	57 948	13.9%
3000–3499	140 775	33.7%
3500–3999	133 305	31.9%
4000+	60 516	14.5%
Gestation (weeks)
<32	5127	1.2%
32–36	26 055	6.2%
37	28 407	6.8%
38	70 563	16.9%
39	115 947	27.8%
40	107 481	25.8%
41+	63 807	15.3%
Year
2012	62 553	15.0%
2013	59 409	14.2%
2014	59 277	14.2%
2015	59 016	14.1%
2016	59 757	14.3%
2017	59 559	14.3%
2018	58 224	13.9%

Table 2.

Characteristics of the cohort (all live births in New Zealand, 2000–2016) by exposure to District Health Boards. Counts have been random‐rounded to be divisible by 3 as per the confidentiality requirements of Stats NZ

	Northland n = 15 210		Waitemata and Auckland n = 96 537		Counties Manukau n = 58 008		Waikato n = 37 440		Bay of Plenty, Tairāwhiti, Lakes and Hawkes Bay n = 50 559		Taranaki, Whanganui, MidCentral and Wairarapa n = 34 470		Capital & Coast and Hutt n = 38 460		Nelson, West Coast and Canterbury n = 56 076		South Canterbury and Southern n = 28 374
	n	%	n	%	n	%	n	%	n	%	n	%	n	%	n	%	n	%
Neighbourhood deprivation (higher = greater deprivation)
Deciles 1–4	2118	13.9%	38 877	40.3%	13 092	22.6%	9177	24.5%	10 329	20.4%	8202	23.8%	16 794	43.6%	23 721	42.3%	12 498	44.0%
Deciles 5 and 6	2340	15.4%	20 925	21.7%	7308	12.6%	7335	19.6%	8454	16.7%	7155	20.7%	6729	17.5%	12 498	22.3%	6027	22.2%
Deciles 7 and 8	3567	23.4%	19 071	19.8%	10 221	17.6%	8880	23.7%	11 889	23.5%	9042	26.2%	6780	17.6%	11 937	21.3%	5934	20.9%
Decile 9	2553	16.8%	8643	9.0%	9858	17.0%	5277	14.1%	7896	15.6%	5055	14.7%	3735	9.7%	4647	8.3%	2583	9.1%
Decile 10	4329	28.5%	8565	8.9%	17 220	29.7%	6516	17.4%	11 496	22.7%	4713	13.7%	4290	11.1%	2529	4.5%	1077	3.8%
Missing	306	2.0%	477	0.5%	333	0.6%	276	0.7%	528	1.0%	324	0.9%	150	0.4%	768	1.4%	270	1.0%
Ethnicity
Asian	642	4.2%	19 827	28.2%	14 094	24.3%	4089	10.9%	3375	6.7%	2373	6.9%	6174	16.0%	6999	12.5%	1929	6.8%
European and other	5049	33.2%	27 978	39.8%	11 214	19.3%	16 278	43.5%	19 377	38.3%	17 334	50.2%	18 333	47.6%	34 833	62.1%	19 206	67.7%
Middle Eastern, Latin American, African	72	0.5%	1932	2.7%	828	1.4%	594	1.6%	381	0.8%	291	0.8%	1017	2.6%	867	1.5%	408	1.4%
Māori	9036	59.4%	11 121	15.8%	14 922	25.7%	14 943	39.9%	25 446	50.3%	13 173	38.2%	8997	23.4%	10 782	19.2%	5763	20.3%
Pacific	414	2.7%	9447	13.4%	16 974	29.2%	1554	4.1%	2010	4.0%	1326	3.8%	3954	10.3%	2619	4.7%	1080	3.8%
Marital status
De facto	5265	34.6%	14 190	20.2%	11 502	19.8%	11 469	30.6%	16 764	33.1%	12 144	35.2%	10 275	26.7%	17 352	30.9%	9981	35.2%
Married	5061	33.3%	46 491	66.1%	31 479	54.2%	17 307	46.2%	19 374	38.3%	14 634	42.4%	21 966	57.1%	30 819	54.9%	14 373	50.6%
Single	4230	27.8%	8631	12.3%	12 495	21.5%	7479	20.0%	12 471	24.7%	6735	19.5%	5490	14.3%	6819	12.2%	3561	12.5%
Unknown	657	4.3%	990	1.4%	2553	4.4%	1206	3.2%	1977	3.9%	975	2.8%	747	1.9%	1110	2.0%	468	1.6%
Maternal age (years)
<20	1233	8.1%	2067	2.9%	3489	6.0%	2304	6.2%	3663	7.2%	2097	6.1%	1398	3.6%	2100	3.7%	1146	4.0%
20–24	3666	24.1%	8157	11.6%	11 808	20.3%	7749	20.7%	10 980	21.7%	7362	21.3%	4974	12.9%	8337	14.9%	4620	16.3%
25–29	4254	28.0%	17 169	24.4%	16 701	28.8%	11 097	29.6%	14 118	27.9%	10 308	29.9%	9045	23.5%	15 351	27.4%	7866	27.7%
30–34	3540	23.3%	24 468	34.8%	15 999	27.6%	10 164	27.1%	13 203	26.1%	9075	26.3%	12 849	33.4%	17 979	32.0%	8940	31.5%
35+	2517	16.5%	18 435	26.2%	10 035	17.3%	6147	16.4%	8622	17.0%	5646	16.4%	10 209	26.5%	12 336	22.0%	5811	20.5%
Maternal smoking
Non‐smoker	10 842	71.3%	65 157	92.7%	46 821	80.7%	30 075	80.3%	38 409	75.9%	27 180	78.8%	34 203	88.9%	48 867	87.1%	23 874	84.1%
Smoker	4137	27.2%	4332	6.2%	7713	13.3%	6903	18.4%	11 775	23.3%	7038	20.4%	3696	9.6%	6825	12.2%	4458	15.7%
Missing	237	1.6%	819	1.2%	3495	6.0%	486	1.3%	405	0.8%	276	0.8%	576	1.5%	408	0.7%	51	0.2%
Maternal education
No qualification	1620	10.6%	3516	5.0%	4995	8.6%	3636	9.7%	4542	9.0%	3438	10.0%	1941	5.0%	4749	8.5%	2283	8.0%
School qualification only	4587	30.1%	15 735	22.4%	15 564	26.8%	10 776	28.8%	14 772	29.2%	11 328	32.8%	9486	24.7%	15 096	26.9%	8247	29.1%
Post‐school qualification below degree	2529	16.6%	10 650	15.1%	10 047	17.3%	6846	18.3%	10 008	19.8%	6462	18.7%	5874	15.3%	10 329	18.4%	5352	18.9%
Tertiary degree	2253	14.8%	23 505	33.4%	9756	16.8%	7932	21.2%	8850	17.5%	6615	19.2%	12 894	33.5%	13 872	24.7%	7383	26.0%
Unknown	4227	27.8%	16 899	24.0%	17 670	30.4%	8268	22.1%	12 411	24.5%	6651	19.3%	8280	21.5%	12 051	21.5%	5121	18.0%
Maternal benefit recipient
No	4968	32.7%	42 129	59.9%	27 582	47.6%	15 957	42.6%	18 285	36.2%	14 160	41.1%	20 841	54.2%	30 534	54.4%	14 514	51.1%
Yes	10 242	67.3%	28 149	40.1%	30 402	52.4%	21 480	57.4%	32 262	63.8%	20 316	58.9%	17 628	45.8%	25 548	45.6%	13 866	48.9%
Number of previous births
0	7314	48.1%	36 696	52.2%	29 082	50.2%	18 531	49.5%	25 755	50.9%	17 268	50.1%	19 935	51.8%	29 097	51.9%	14 439	50.9%
1	4191	27.5%	22 620	32.2%	16 347	28.2%	11 109	29.7%	14 592	28.8%	10 422	30.2%	12 477	32.4%	17 940	32.0%	9219	32.5%
2	2085	13.7%	7476	10.6%	7047	12.2%	4923	13.1%	6417	12.7%	4407	12.8%	4326	11.2%	6390	11.4%	3456	12.2%
3+	1629	10.7%	3513	5.0%	5553	9.6%	2895	7.7%	3822	7.6%	2394	6.9%	1737	4.5%	2673	4.8%	1272	4.5%
Sex of infant
Female	7458	49.0%	34 044	48.4%	28 254	48.7%	18 234	48.7%	24 645	48.7%	16 734	48.5%	18 732	48.7%	27 330	48.7%	13 851	48.8%
Male	7755	51.0%	36 261	51.6%	29 775	51.3%	19 227	51.3%	25 941	51.3%	17 760	51.5%	19 743	51.3%	28 773	51.3%	14 535	51.2%
Birthweight (g)
<2500	795	5.2%	4044	5.8%	3645	6.3%	2262	6.0%	3168	6.3%	2142	6.2%	2292	6.0%	3177	5.7%	1587	5.6%
2500–2999	2052	13.5%	10 524	15.0%	8790	15.2%	5028	13.4%	7071	14.0%	4458	12.9%	5106	13.3%	7077	12.6%	3432	12.1%
3000–3499	4971	32.7%	25 131	35.7%	19 950	34.4%	11 973	32.0%	16 803	33.2%	11 148	32.4%	12 948	33.7%	18 462	32.9%	9168	32.3%
3500–3999	4914	32.3%	21 693	30.9%	17 493	30.2%	12 117	32.4%	16 233	32.1%	11 394	33.1%	12 432	32.3%	18 642	33.2%	9612	33.9%
4000+	2469	16.2%	8883	12.6%	8130	14.0%	6054	16.2%	7275	14.4%	5316	15.4%	5661	14.7%	8730	15.6%	4572	16.1%
Gestational age (completed weeks)
<32	162	1.1%	747	1.1%	783	1.4%	474	1.3%	597	1.2%	501	1.5%	462	1.2%	696	1.2%	333	1.2%
32–36	807	5.3%	4146	5.9%	3720	6.4%	2256	6.0%	3309	6.5%	2286	6.6%	2517	6.5%	3555	6.3%	1893	6.7%
37	849	5.6%	4908	7.0%	4347	7.5%	2430	6.5%	3414	6.8%	2271	6.6%	2703	7.0%	3645	6.5%	1818	6.4%
38	2388	15.7%	12 798	18.2%	11 142	19.2%	5625	15.0%	8613	17.0%	5205	15.1%	6222	16.2%	8535	15.2%	4644	16.4%
39	3909	25.7%	19 995	28.4%	16 443	28.4%	9648	25.8%	13 653	27.0%	9339	27.1%	10 551	27.4%	16 143	28.8%	7845	27.6%
40	4476	29.4%	17 622	25.1%	13 932	24.0%	9801	26.2%	13 089	25.9%	9174	26.6%	10 089	26.2%	14 982	26.7%	7308	25.8%
41+	2622	17.2%	10 077	14.3%	7632	13.2%	7206	19.2%	7878	15.6%	5712	16.6%	5919	15.4%	8526	15.2%	4536	16.0%

There were 267 SUDI cases, giving an overall rate of 0.64/1000 live births during the study period (2012–2018). The SUDI rate varied from 1.11/1000 in Northland to 0.30/1000 in Waitemata and Auckland. Counties Manukau DHB had the largest number of deaths (n = 54; rate = 1.08/1000). Table 3 shows the unadjusted RRs and their 95% CIs. Compared with the reference group, Waitemata and Auckland DHBs, the unadjusted RR for Northland was 2.67 (95% CI = 1.49, 4.79), Counties Manukau DHB 2.36 (1,56, 3.58), Waikato 2.10 (1.31, 3.38), Taranaki, Whanganui, MidCentral and Wairarapa 1.99 (1.21, 3.26) and Bay of Plenty, Tairāwhiti, Lakes and Hawke's Bay 1.86 (1.18, 2.92), respectively. However, after adjustment, no DHB had a significantly increased risk of SUDI. The geographic variation is shown graphically in Figure 1.

Table 3.

The cohort and SUDI cases by District Health Board and their unadjusted and adjusted RR (95% CI). Counts (n columns) have been random‐rounded to be divisible by 3 as per the confidentiality requirements of Stats NZ

				Unadjusted			Adjusted†
	Cohort	SUDI (n)	SUDI rate per 1000	RR	95% CI		RR	95% CI
District Health Board
Northland	15 210	15	1.11	2.67	1.49	4.79	1.10	0.60	2.02
Waitemata and Auckland	96 537	39	0.30	Ref			Ref
Counties Manukau	58 008	54	1.08	2.36	1.56	3.58	0.99	0.64	1.52
Waikato	37 440	33	1.01	2.10	1.31	3.38	1.11	0.68	1.82
Bay of Plenty, Tairāwhiti, Lakes and Hawke's Bay	50 559	39	0.59	1.86	1.18	2.92	0.78	0.49	1.26
Taranaki, Whanganui, MidCentral and Wairarapa	34 470	27	0.86	1.99	1.21	3.26	1.12	0.67	1.87
Capital & Coast and Hutt	38 460	18	0.54	1.32	0.77	2.27	1.19	0.69	2.05
Nelson Marlborough, West Coast and Canterbury	56 076	27	0.38	1.31	0.81	2.13	1.38	0.84	2.27
South Canterbury and Southern	28 374	15	0.59	1.34	0.74	2.44	1.33	0.72	2.47

CI, confidence interval; RR, relative risk; SUDI, sudden unexpected death in infancy.

^†Adjusted for: birthweight, gestation, sex of infant, ethnicity, marital status, maternal age, number of previous births, maternal smoking, neighbourhood deprivation, maternal benefit recipient, maternal education and year.

Fig. 1.

Regional variation of SUDI mortality by District Health Boards in New Zealand, 2012–2016. (a) shows the unadjusted relative risks compared with Waitemata and Auckland and (b) shows the relative risks after adjustment for individual‐level confounders. Blue represents above average and red below average relative risk. SUDI, sudden unexpected death in infancy.

The covariate results are also of interest (Table 4). In the adjusted analysis, the most deprived decile was associated with an increased risk. Māori and Pacific infants, single mothers, young maternal age and maternal smoking were all associated with SUDI. Compared to first‐born infants, there was no increased risk of being second born, but children born third or fourth or later had increased risk. Boys and infants born preterm (32–36 weeks) were also at higher risk of SUDI.

Table 4.

The adjusted RR (95% CI) of the confounders and covariates

	Adjusted†
	RR	95% CI
Neighbourhood deprivation (higher = greater deprivation)
Deciles 1–4	Ref
Deciles 5 and 6	1.12	0.66	1.93
Deciles 7 and 8	1.16	0.71	1.91
Decile 9	1.22	0.72	2.07
Decile 10	1.82	1.12	2.95
Unknown	1.66	0.51	5.42
Ethnicity
Asian	0.89	0.43	1.84
European and other	Ref
Middle Eastern, Latin American, African	Undefined
Maori	1.55	1.06	2.26
Pacific	2.16	1.36	3.43
Marital status
Married	Ref
De facto	1.50	0.95	2.36
Single	2.14	1.34	3.41
Unknown	3.96	2.10	7.44
Maternal age (years)
<20	2.29	1.43	3.67
20–24	1.77	1.19	2.62
25–29	1.09	0.72	1.63
30–34	Ref
35+	0.70	0.41	1.17
Maternal smoking
Non‐smoker	Ref
Smoker	1.78	1.34	2.36
Unknown	1.25	0.59	2.62
Maternal education
No qualification	1.86	0.92	3.75
School qualification only	1.59	0.83	3.06
Post‐school qualification below degree	1.12	0.55	2.31
Tertiary degree	Ref
Unknown	1.75	0.90	3.39
Benefit recipient
No	Ref
Yes	1.43	0.97	2.10
Number of previous births
0	Ref
1	1.22	0.84	1.76
2	2.31	1.54	3.48
3+	2.33	1.43	3.78
Sex of infant
Female	Ref
Male	1.30	1.01	1.66
Birthweight (g)
<2500	1.01	0.59	1.71
2500–2999	1.11	0.77	1.58
3000–3499	Ref
3500–3999	0.92	0.65	1.29
4000+	0.65	0.39	1.08
Gestation (weeks)
<32	0.25	0.11	0.61
32–36	1.68	1.00	2.84
37	1.32	0.80	2.17
38	1.42	0.97	2.10
39	0.95	0.65	1.38
40	Ref
41+	0.79	0.48	1.30
Year
2012	Ref
2013	1.29	0.83	2.02
2014	1.18	0.76	1.84
2015	1.10	0.70	1.73
2016	1.25	0.80	1.95
2017	1.27	0.81	1.98
2018	0.64	0.37	1.10

CI, confidence interval; RR, relative risk.

^†Adjusted for all the variables in the table and district health board.

Discussion

As expected, SUDI risk varied with DHB. However, when other factors were considered DHB was not significant. This supports, in part, the contention ‘Many factors influence SUDI rates in DHBs, including service delivery and the characteristics of the population the DHB serves’. ⁷ We were unable to measure service delivery and, indeed there is no consensus on the definition and indicators of the quality of the health‐care services. ²² , ²³ However, we would argue that there is no evidence to support the statement that variation in service delivery is influencing regional SUDI rates in New Zealand as the differences seen can be explained by socio‐economic and demographic differences.

Many of the confounding variables cluster together, for example, DHB, young maternal age, single, Māori and Pacific ethnicity, neighbourhood deprivation, maternal smoking and benefit recipient. These are important factors which are associated with SUDI and are outside the role of health services. These social determinants of health have an important influence on health inequities seen within and between countries. At all levels of income, health and illness follow a social gradient: the lower the socio‐economic position, the worse the health. ²⁴ This has been recognised by the Ministry of Health's Expert Advisory Group on SUDI Prevention. ²⁵ Their first recommendation was the recognition that relief from poverty was a fundamental measure in the prevention of SUDI deaths.

Māori and Pacific infants were at higher risk of SUDI than European infants. We have previously shown from two case–control studies that the higher rate in Māori is due to a higher prevalence of risk factors for SUDI among Māori than non‐Māori, especially smoking. ²⁶ , ²⁷ It is likely that bed sharing explains the residual higher risk for Māori and Pacific infants in New Zealand.

The major limitation of this study is the small number of SUDI cases, such that we may have made a type 2 error, that is, we have concluded that there is not a significant effect, when there really is. However, including earlier years of data would have introduced changing mortality rates, which cannot be explained by the variables examined in this analysis. Furthermore, the analysis was not able to adjust by important risk factors, such as sleep position and bed sharing.

Our study is consistent with a recent study from the United States. ²⁸ There was marked geographic variation in SUDI rates by state, but the variation was attenuated after adjusting for covariates including known risk factors for SUID.

Conclusion

Our study has shown that there is marked variation in SUDI risk by DHB, but this is predominantly explained by socio‐economic and demographic variation within DHBs. This study emphasises the important contribution of social determinants of health to SUDI. We support the call to address poverty. The focus should be on families in poverty rather than specific DHBs, with culturally appropriate resources allocated to DHBs with high needs.

Ethical approval

All the data were non‐identifiable. This study was approved by the University of Auckland Human Participants Ethics Committee (UAHPEC, reference number 019714).

Data availability statement

The data used in this study are held with the Integrated Data Infrastructure and are managed by Statistics New Zealand. These data are publicly available, although access is restricted. Please see https://www.stats.govt.nz/integrated-data/integrated-data-infrastructure/ for more details.