Adversarial medicolegal cases in the UK involving children usually with severe birth injury leading to severe functional impairment in cerebral palsy (CP) need life expectancy estimates by legally instructed medical experts. These estimates are pertinent because, in cases with the same facts and background data, most commonly Brooks et al.,¹ life expectancy estimates have been materially different, resulting in potentially marked differences in what lump sum payments might be awarded by the court.

As a result, lawyers acting in the best interests of their clients will often opt for the highest (claimant side) or lowest (defence side) life expectancy that the experts provided. This is almost inevitable in any adversarial system and the experts are in the middle.

Being a respiratory paediatrician and not a neurologist, nevertheless I am asked to be involved in life expectancy estimates in those children with the most severe impairments either to: (1) estimate life expectancy; (2) opine as to whether there should be a respiratory adjustment of other experts’ estimates of life expectancy, or (3) comment on the ‘other side’s’ life expectancy estimates. This results in an interest in how life expectancy is, could, or should be estimated, whether life expectancy is appropriate at all, and what, if any, adjustments for comorbidities such as epilepsy, scoliosis, and cognition ought to be considered. In addition, issues of the relevance of general population life expectancy within and between countries and therapeutic advances also need consideration. As my experience is with only the most severely impaired children, they will form the focus of my discussion.

METHODOLOGY

Brooks et al.¹ do not use Gross Motor Function Classification System (GMFCS) levels, but a functional categorization, which while inevitably related to the GMFCS, is not the same. Indeed, given that GMFCS level V includes children who are ‘limited in their ability to maintain antigravity head and trunk postures and control leg and arm movements’, children in Brooks et al.’s most severe tube-fed, cannot lift head when prone (TF/CLH) category would, in general, be even more severely impaired than children in GMFCS level V, which may be an argument for additional GMFCS categories.

Life expectancy is defined as the mean survival time remaining for a cohort or the average for an individual represented by that cohort if theoretically that individual’s life could be lived repeatedly. The ‘normal’ population age at death has a large left skew, i.e. numbers of deaths per year of life after infancy are very low until about age 50 years when they markedly rise to peak in the 80s and then rapidly fall to effectively zero at around 115 years. Sir David Spigelhalter, Professor of the Public Understanding of Risk at Cambridge University, has criticized life expectancy as a ‘misleading summary’.² From typical UK population life expectancy tables 2010 to 2012, he states that while the male mean life expectancy at birth is 79 years, the median is 82 years. The most common age at death, the mode, is 86 years. So, for left skew distributions this is the outcome and for right skew distributions, such as income, the opposite will be true, with 2019 UK mean disposable income³ £35 000, median £29 000, and the mode only £20 000.

Brooks et al.’s¹ data provides survival probabilities for children under 15 years of age adjusted to year 2010 and life expectancy after that (Table 1 and Fig. 1).

Day et al.,⁴ using Strauss’ earlier data,⁵ show that median survival of children aged 4 years in the most severe TF/CLH category was 17 years, but life expectancy was 20 years, nearly 18% greater, demonstrating a significant right skew. Indeed, survival of such children is an example of exponential decay, with a large right skew. One can use general population data, and assumptions about the relationship between a population with CP and the general population, to extrapolate beyond the known CP survival data using fairly complex mathematical formulae.⁶

So, given that survival times are skewed to the left in the general population and the available evidence on survival of children with severe CP shows the opposite to be true, then in scientific work, when distributions are found to be significantly skewed, reporting medians rather than means is generally required. In light of this, UK courts may want to revisit their position of focusing on life expectancy (mean survival time) when awarding damages in such cases. In addition, in these severe categories, regardless of the type of data used, be it period, projected, or cohort, median survival will occur considerably earlier than the mean, making the median more up-to-date, a superior measure of central tendency less affected by outliers, and compared with the mean, rarely requires assumptions or mathematical manipulation to determine it.⁶

TYPICAL POPULATION LIFE EXPECTANCY DIFFERENCES BETWEEN COUNTRIES

Nevertheless, many UK expert’s reports refer to the whole of USA life expectancy rates (in 2018: 78y 8mo)¹⁰ and so propose an increase in life expectancy of 2 years because the child is in the UK, which cannot be correct, or as a proportion, which would make a small difference.

PROJECTED LIFE EXPECTANCIES INTO THE FUTURE

An argument often put forward is: if ffrom now there are improvements in general life expectancy into the future, surely a child in Brooks et al.’s TF/CLH category, for example, would be expected to benefit from a proportion of that? From this, two questions arise: first, will there be a continued rise in general life expectancy into the future, with obesity, pandemics, climate change, diabetes, antibiotic resistance, etc., all having the potential to halt or even reverse this. Second, if there are improvements, how likely is it that those in the most severe categories (GMFCS levels IV–V or TF/CLH) would be expected to benefit? (note, therapy-induced advances in life expectancy are dealt with below). Dealing with the first question, recent history has shown that although UK (and US) life expectancy markedly increased by some 10+ weeks per year from 1980¹¹ until 2010, there was then a marked fall, reaching nearly zero in 2016 in the UK with a similar pattern in the USA.¹² Indeed, life expectancy declined in 2020, which exemplifies the risks of extrapolating life expectancy rates from historical data.

Whether these past improvements would apply to children in the TF/CLH category is problematic, as changes in general life expectancy in the age ranges that children in the TF/CLH category usually survive to have been tiny compared with infancy and those greater than 50 years of age. In the general UK population in 2018 (Fig. 2), 98% of males and 99% of females already survive the 36 years from age 4 to 40 years with almost no room for improvement and virtually no change since 1998, whereas since 1998 again over 36 years but now from age 40 to 76 years, survival rose 26% in males and 12.5% in females.¹³ Given the largest cause of death in young adults is suicide and poisonings¹⁴ followed by ischaemic heart disease, malignancies, and trauma, it is very doubtful, in my view, that if all suicides and poisonings etc., could be prevented, it would improve the life expectancy of a child in the TF/CLH category. Changes in the elderly similarly are very unlikely to be relevant in the most severe categories, but would be relevant in milder cases and, indeed, improvements in children in milder disability categories match the change in the general population.¹⁵ Therefore, in my opinion, general life expectancy changes should not be used in the most severe cases.

THERAPY-INDUCED ADVANCED LIFE EXPECTANCY

Over the last decades, treatment and care for children with severe CP has undeniably improved, with added focus on the management of seizures, nutrition, muscle tone, scoliosis and joints, infection, and respiratory issues. Respiratory management has contributed through the now widespread availability of vaccinations, cough assist machines, routine chest physiotherapy, and the better management of secretions and aspiration. Non-invasive ventilation is now widely available in specialist centres, but probably insufficiently used in children with severe CP. Tracheostomy, inevitably highly contentious, may well increase survival in a severely affected individual compared with the same child/adolescent who does not receive one, but ignores quality of life considerations. Whilst all the above have undoubtedly improved survival, this is likely to plateau because these benefits assist with the consequences of severe CP rather than the brain injury/CP itself. In addition, further treatment of symptoms will not have the impact on survival compared with, for example, a single dose of onasemnogene abeparvovec in spinal muscular atrophy type 1.

IMPACT OF SEIZURE DISORDERS ON LIFE EXPECTANCY IN SEVERE CP

It is recognized that poor seizure control increases the risk of unexpected death.¹⁶ However, for the most severe categories, the issue is how much the seizure disorder is already ‘baked into’ measures of survival, given that the Australian Cerebral Palsy Register¹⁷ found that 65% of children in GMFCS level V at age 5 years had a seizure disorder. In Taiwan,¹⁸ there was a seizure prevalence over all GMFCS levels of 30%, but with no subdivisions. My personal experience is that more than 95% of all children in the TF/CLH category seen for medicolegal reasons had a diagnosed seizure disorder. If the considerable majority of the most severely affected children have a seizure disorder, then Brooks et al.’s survival data for the TF/CLH category very likely reflects that.¹ So rather than remove 10% from life expectancy in such children with a seizure disorder, as is frequently done, perhaps 10% should be added for the seemingly rare event of a child in the TF/CLH category without a seizure disorder. In the Victoria, Australia data¹⁹ (see below), epilepsy’s risk on life expectancy was the least of the six categories cited.

IMPACT OF COGNITION ON LIFE EXPECTANCY FOR A GIVEN SEVERITY CATEGORY

Consider a male of 8 years who sits unsupported and partially rolls, is fully gastrostomy fed, vocalizes without speech, but has sufficient cognition to scroll and use, to a degree, a tablet with his feet and may be able to count to three. He was classed by others to be in GMFCS level IV, but the key issue was his cognition. The UK,²⁰ South of Sweden,^{21, 22} California,¹ and Okinawa, Japan^{23, 24} cohorts do not use cognition as a specific feature. The West of Sweden cohort^{25, 26} showed cognition was significant in a multivariate but unadjusted model, however the Victoria¹⁹ and Western Australia²⁷ CP cohorts went further. In Victoria, severe/profound cognitive impairment was an independent risk factor leading to 3.01 (95% confidence interval 1.74–5.22) times the risk of death compared with no impairment. This single result appears to come from calculations across all GMFCS levels and while the risk was raised, its magnitude was not provided. In Western Australia, cognition (IQs 50–69, 35–49 and <35) forms part of the Western Australian disability score (DISAB), comprising a maximum score of 3 out of a possible 12. The data provided at least allows one to produce Table 2. Table 2 shows a very unlikely cliff edge in life expectancy in that life expectancy differences are the same for children with DISAB scores of 6 and 12 as they are for DISAB scores of 8 and 9.

Thus, in the real world example mentioned earlier, the 8 year-old male was scored by others as 3 and 3 for his motor impairments and nil for epilepsy, blindness, and deafness making 6. Therefore, the result depended on whether his cognition score was 2 (IQ 35–49) or 3 (IQ<35) because from Table 2 a DISAB score of 9 at age 8 years gives an additional 33 years 10 months (total 42y) of life expectancy. This is almost identical to what would be derived from Brook et al.’s paper. However, if changing the DISAB score to 8, which theoretically could be a 1-point change in IQ, and which is notoriously difficult to measure in such children, then the Western Australia data now suggests a life expectancy of 8 years+55 years 4 months = 63 years 4 months. The Victoria cohort was further reviewed regarding cognition,^{28, 29} finding that non-ambulant children, such as in the example above, born between 1981 and 2010 (n=3248) with an IQ <70, so including the current definition of mild intellectual impairment, only had a 44% chance of reaching 35 years; however, this included everyone from birth, not from 4 years, as in Brooks et al.’s data.¹ From the above example, given the life expectancy estimates now range from 35 to 63 years, it requires no imagination to realise which predicted outcome each side’s lawyers would prefer. The Western Australian authors freely acknowledge that the DISAB score needs refinement. In this regard it would be an ideal opportunity to use the Western Australian data to formulate a mathematical model to predict life expectancy or median survival for the severe forms of disability and then evaluate it on the Victoria data and vice versa. This could determine whether blindness, as an example, used in the DISAB score is actually relevant as it is non-significant in the Victoria cohort. If Victoria data’s independent adjusted risks for life expectancy were ranked, they would be severe movement disorder (6), severe or profound intellectual impairment (3), deafness (2.6), lack of speech (2.5), term birth (1.8), and epilepsy (1.4).

THE IMPACT OF SCOLIOSIS AND ITS TREATMENT ON LIFE EXPECTANCY

Scoliosis is an important aspect of respiratory health, given how many of the most severely affected children develop it. All but one of the cohorts cited above do not mention scoliosis at all. However, the South of Sweden cohort now has incidence/prevalence data³⁰ stating that 20% of children in GMFCS level V have a severe scoliosis with a Cobb angle >40 degrees when aged 10 years and 75% at 20 years, rising because of the effects of puberty, but no specific data on whether life expectancy is altered as a result. However, as with epilepsy, given its frequency in children in the TF/CLH category, it would be the comparatively rare child without a severe scoliosis that may well merit consideration of being ‘above average’ with, perhaps, a raised estimate of survival. While scoliosis surgery for the relief of pain or to aid seating would, anaesthetic risk permitting, seem incontrovertible, for ‘respiratory benefit’, I am far more dubious. Indeed, the literature for idiopathic scoliosis at most only shows very modest benefit of surgery.^{31, 32}

So where does that leave the hapless ‘expert’? Well, we could admit that, in estimating survival in whatever form, we are largely trying to divide a cow by a tree and factorizing the answer because applying group data to any individual is fraught with potential error and inevitably will be wrong.

In view of this the following are my conclusions. (1) Median survival is far more likely to be a much better reflection of the likely outcome than mean life expectancy. (2) There is no need to adjust for California life expectancy if using Brooks et al.’s data on UK patients. (3) There is no current justification for adding future life expectancy changes as they will mostly continue to apply to infancy and those over 40 years and far less in between. (4) There is no radical ‘curative’ therapy on or indeed beyond the horizon, and management improvements will make only small differences. However, their more widespread application may make a larger difference. Tracheostomies deserve further debate and data acquisition. (5) Given their frequency in GMFCS level V and TF/CLH categories, epilepsy and scoliosis, unless absent for epilepsy or mild for age for scoliosis, are largely accounted for in survival calculations. (6) The key imponderable is the role of cognition in life expectancy. Is it already accounted for in the severest disability categories and may it be key in milder physical disability categories? As a pulmonologist, this is definitely not my specialism, but to avoid endless cherry picking by lawyers and the pressuring of experts to revise/adjust/‘shave’ their opinions to avoid being grilled in court, I urge further discussion. (7) Of course, no fault compensation would avoid some needless combat, but as I understand it, in Australia in no-fault settings, adversarial legal redress is still frequently sought.

Acknowledgements

I am grateful to Dr Ian Balfour-Lynn and Professor Andrew Bush for their helpful comments from a respiratory viewpoint in the draft stages and the assistance of Dr Ben Lloyd, paediatric neurologist with whom I have sat in many a conference and occasionally disagreed with. The author has stated that he had no interests that might be perceived as posing a conflict or bias.