By Ellen Stewart

Patient and public involvement (PPI) in health research is a field of practice which has grown at pace, and is now understood as central to high quality health research. The NIHR defines public involvement in research as “research being carried out ‘with’ or ‘by’ members of the public rather than ‘to’, ‘about’ or ‘for’ them”. This imperative also applies to public health research. For example, the NIHR’s Public Health Research programme specifies a requirement for public involvement (rather than the more conventional NIHR framing of public and patient involvement) and directs researchers to the standard NIHR INVOLVE guidance and resources.

This shift in focus requires researchers to reflect carefully on the processes they put in place. Public health almost always operates with a population lens, and the knowledge sought out in the form of ‘lived experiences’ are much more widely distributed than in the examples of, say, respiratory medicine. The parallel in that field is that a public health lens might centre all our experiences of living with excessive levels of air pollution, while a healthcare lens would reasonably focus on accessing the knowledge of people with experience of living with a specific respiratory health condition.

Sometimes this wider lens, and lack of an obvious patient community to recruit through, means that public health research can be quite poor at engaging with communities. In large-scale modelling projects, like SIPHER, this can be magnified because the data we work with is far removed from the people whose lives it reports.

We wanted SIPHER to do something different, and our approach has a couple of distinctive elements.

One is that engagement is a core part of the consortium, sitting within the overall Consortium Management Group in a workstream on evaluation, and delivered by a Co-Investigator on the grant. The goal here is that engagement is integrated into the intellectual agenda of SIPHER, within a birds-eye view of the complex and multi-faceted work of the consortium across many organisations, rather than operating separately within each workstream.

The second is that, rather than set up select panels with people we recruit directly from a defined population of patients, we work in partnership with local community organisations in the geographical area of each policy partner. These organisations bring a range of strengths to the consortium and aren’t simply recruiting participants for our panels. They have significant place-based local knowledge about the publics they work with and serve. Staff have existing trusting relationships with service users and local community members, which should facilitate our panel members to inform and challenge our work from the vantage point of their own knowledge of the issues. These collaborations mean we can direct the budget for venue hire, catering and facilitation into organisations who are working for these local communities, and who are in many cases struggling to keep afloat due to the wider context of austerity.

Across the remaining years of the programme, we hope that these community partner organisations will scrutinise and shape our work from the perspective of people disadvantaged by the kinds of inequalities we will be modelling.

Much of the last six months has been spent planning ways to move workshops with our community panels online, paying particular attention to making workshops digitally inclusive and accessible. This should ensure we still enable our participants and partner organisations to engage with the research, even if we aren’t sitting round a table together with tea, biscuits and a lot of post it notes. We will post regular updates on our community panel workshops across the remaining years of SIPHER’s work.

By Colin Angus

The COVID-19 pandemic has had an almost unprecedented impact on mortality across the globe and the United Kingdom has been one of the worst affected countries, with an estimated 53,937 more deaths than expected in England and Wales alone in the first half of 2020. Recent data from the Office for National Statistics somewhat surprisingly suggested that the prevalence of COVID-19 in early September was highest among the least deprived groups in England. Some people have drawn parallels in these findings to the Spanish flu pandemic in 1918-19 whose first wave overwhelmingly affected poorer groups, while the second wave had a greater impact on more affluent parts of society. However, this comparison assumes that the first wave of COVID-19 actually did have a greater impact on more deprived groups. What does the data have to say about this?

If we look at the raw numbers for England and Wales, it’s clear that there were slightly more deaths from all causes between 1^st March and 31^st July among people living in more deprived areas. This was particularly the case for deaths that were confirmed as COVID-19 on the death certificate, of which there were around a third more in the most deprived, compared to the least deprived decile of the population. The inequality in deaths from other causes over the same period is somewhat smaller. This provides limited support for the idea that the most deprived areas were hit hardest by the pandemic, at least in terms of deaths.

But, simply comparing the numbers of deaths obscures something important about what is going on. We are measuring deprivation here using deciles of the Index of Multiple Deprivation, which means that 10% of the English population falls into each decile, but there is significant variation in the age distribution of each decile. People living in the most deprived areas are much more likely to be younger, and while they may be in poorer health (health is one component of how deprivation is calculated), we know that older ages are at significantly greater risk from many health conditions, including COVID-19.

As a result, it makes sense to account for these differences when comparing between deprivation groups. This is typically done through a process called ‘age-standardisation’ whereby we estimate how many deaths there would have been in each decile if the age distribution of each population was the same.

If we age-standardise the numbers from the first graph above then we see a much more striking level of inequality. Based on the available data, if the age structures were the same, we would expect to have seen over double the number of deaths in the most deprived decile compared to the least deprived. Again, the inequality gradient is steeper in deaths from COVID-19 than from other causes, but there is also a substantial inequality in age-standardised deaths from non-COVID-19 causes.

Is this our answer? Can we now definitively say that the mortality burden of COVID-19 has fallen disproportionately on the most deprived groups?

Well, no. Because, as sad as it is to say, deaths are always unequal. Life expectancy for people living in the most deprived 10% of areas in England is considerably shorter than for people living in the least deprived 10% – 9.5 years shorter for men and 7.7 years for women. Therefore, while deaths in 2020 might be unequal, the important question is really whether they are more unequal than usual.

When we compare deaths by deprivation decile between 1^st March and 31^st July in 2020 to deaths from the years 2014 to 2018 (shown as dark blue dots), the most obvious difference is the larger number of deaths this year across all decile. However, we can also see that the difference between the cluster of dots and the top of the bars (i.e. the excess deaths in 2020 over 2014-18) is greater in the more deprived groups.

Before we use these figures to definitively answer our question, we first need to decide how we are measuring inequality. There are many different ways of measuring inequality in deaths, but these tend to fall into two broad categories:

• Measures of absolute inequality – these reflect the size of a difference between groups in the population
• Measures of relative inequality – these reflect the proportional difference between groups in the population

Measures of both types are widely used. Which is the most appropriate for any given situation depends on your perspective. For example, deaths from cardiovascular disease have fallen rapidly in many countries in recent decades, but these falls have been largest in the least deprived groups. As a result absolute inequality has fallen, while relative inequality has increased.

Two of the most commonly used measures of inequality are the Slope and Relative Indices of Inequality – SII and RII respectively. Broadly speaking, the SII is the absolute difference between the most and least deprived groups and the RII is the relative difference (i.e. the ratio between them). The above graph shows clearly that the SII in 2020 is greater for both men and women in 2020 – by this measure absolute inequality has risen by 30% in both men and women from 2018 levels. Relative inequality has also increased, but only very slightly. The coloured dots in the graph below illustrate that inequality in deaths from COVID-19 is greater than ‘usual’, while inequality from other causes is similar to, or perhaps slightly lower than in previous years.

What does this mean for our original question? Is it true to say that the first wave of COVID ‘Hit the poor’? As ever, it’s complicated, but ultimately, while it is true to say that there have been more deaths during the first half of 2020 in more deprived groups, this is merely a continuation of a historical trend.

So far we have only looked at data from England – do we see a different story when we look at Scotland? It is certainly true that there were substantially more confirmed COVID-19 deaths in the most deprived quintile in Scotland (note that there are some minor differences in the way that deprivation is defined in Scotland compared to England).

In fact, when we also look at the raw number of deaths from both COVID-19 and other causes between 1^st March and 31^st August 2020, we can see markedly bigger inequalities for both when compared to the equivalent graph for England (right back up at the top of the blog).

However, once we age-standardise and compare 2020 to recent pre-COVID years, the pattern looks very similar to the one we see in England – absolute inequality is higher in 2020, while relative inequality is at a similar level to previous years. The only real difference between England and Scotland is that relative inequality has been considerably higher in Scotland for several years – in 2020 the RII is 2.4 for men and 2.2 for women compared to 1.6 and 1.5 respectively in England.

Overall, it is probably true to say that COVID-19 has perpetuated existing inequalities in mortality more than that it has created new inequalities. There have been more deaths in more deprived groups during the pandemic, but this inequality is not a new phenomenon. Similarly, inequality in deaths has been greater in 2020 in Scotland than in England, but this is not a new phenomenon either.

Of course, the analysis here concentrates on only one outcome, while the pandemic will have impacts across many aspects of society. It is very possible that the direct and indirect impacts of COVID-19 will lead to a huge range of inequalities. An important challenge for the SIPHER project will be to examine and understand some of these diverse impacts and explore how they interact with each other to influence health and society – watch this space.

All the analysis in this blog post was conducted using R open source statistical software and all code can be found here, here, here and here.