Mortality in England – erroneous attribution of excess winter deaths to underlying trend

Hiam et al.’s recent article in your journal drew widespread attention for its perspective on mortality in England and Wales in recent years. ¹ In the opening paragraph of results, the paper states that: ‘ Figure 1 shows that, following many years of decline, albeit with some year-to-year fluctuations, the downward ASMR trend does reverse after 2011 … These figures are age-standardised using 5-year age groups’.

Figure 1.

Age-standardised death rate per 100,000 population, England and Wales, 1980–2015.

However, examination of the presented data demonstrates clearly that these are crude rather than age-standardised rates. I enclose for comparison the original graph and the crude mortality rates for England and Wales over the period 1995–2015 using the same scales, which are manifestly the same data (Figure 1).^2,3

In contrast, the directly age-standardised mortality rates for the period 1995–2014 are compared with crude rates for the same period (Figure 2).^4,5

Figure 2.

Directly standardised mortality rate per 100,000 population, England and Wales 1995–2015 compared with crude mortality rate for the same period.

The argument that mortality rates reversed after 2011 is less persuasive when considered in the light of direct age-standardisation, and the visual appearance of a rise is heavily dependent upon the 2015 data point.

The paper also states that: ‘[Age-standardised mortality rate] … by 2015 … was higher than in any year since 2008 and was 4.8% higher than in 2014 and 2.8% higher than in 2013 … These figures are age-standardised using 5-year age groups’.

In fact, the quoted figures are only correct when comparing crude rates for England and Wales. The directly standardised rate for England and Wales was actually higher in every year up to and including 2010 – the 2015 rate was 4.5% higher than 2014 and 1% higher than 2013.

The excess of winter deaths in 2014–2015 – which hit a high of 43,850 for England in Wales in comparison to 17,310 in 2013–2014 – fell in 2015–2016 to 24,300, a figure lower than the average of the preceding 10 years. ⁶ This suggests that although the 2015 age-standardised mortality was certainly high when compared to trend, it will be very much closer to trend for 2016 data when they become available.

Figure 2 of the paper illustrates the percentage increase in death rates at individual ages of 75 to 90 years from 2014 to 2015 for England and Wales. While it is certainly true that the mortality rates for those individual ages were higher in 2015 than 2014, I have been unable to reproduce the percentage increases illustrated in Hiam et al.’s paper. The data sources cited in the paper appear only to provide single-year population estimates to the age of 89, with older estimates aggregating 90+ years. Replicating the described approach, I can demonstrate a rise of only about half that illustrated in the original paper. For comparison, the Office for National Statistics figures are tabulated (Table 1).

Table 1.

England and Wales, age-specific deaths, population estimates, rates and percentage increase 2014–2015.

	Age-specific deaths		Population estimate		Rates
Age	2014	2015	2014	2015	2014	2015	% Increase
75	11,164	11,208	416,017	410,486	2683.54	2730.42	1.75
76	11,884	12,344	403,059	404,720	2948.45	3050.01	3.44
77	12,440	13,024	380,522	390,576	3269.19	3334.56	2.00
78	13,068	13,628	360,032	367,618	3629.68	3707.11	2.13
79	13,950	14,588	339,937	346,269	4103.70	4212.91	2.66
80	14,699	15,663	312,287	324,938	4706.89	4820.30	2.41
81	15,450	16,156	294,490	296,883	5246.36	5441.87	3.73
82	16,672	17,234	282,672	278,088	5898.00	6197.32	5.07
83	17,778	18,535	264,809	264,593	6713.52	7005.10	4.34
84	18,261	19,520	242,927	245,732	7517.07	7943.61	5.67
85	18,355	19,976	215,312	223,300	8524.84	8945.81	4.94
86	18,274	19,849	190,230	195,581	9606.27	10,148.74	5.65
87	18,566	19492	170,093	170,607	10,915.21	11,425.09	4.67
88	18,191	19,568	150,189	150,400	12,112.07	13,010.64	7.42
89	17,649	18,996	129,328	130,672	13,646.70	14,537.16	6.53

All data are for persons in England and Wales.

By applying the derived 2014 rates to 2015 population estimates, I find a figure for excess deaths of 21,231 for all ages, or 20,359 for ages 75 and over. In both instances, the 90+ deaths are aggregated and expressed as a rate using the 90+ population estimates as a denominator. This excess of deaths would be close to that reported for the high 2014–2015 excess winter deaths figure cited above, but falls considerably short of the 33,798 quoted in the paper.

Hiam et al.’s paper was widely reported in the press, but the data presented were incorrectly described, and the apparent rise since 2011 that they illustrate appears to have been over-interpreted.