Discrepancies in the estimates of life expectancy after SCI

Middleton et al.¹ recently reported life expectancies after spinal cord injury (SCI), based on a cohort of 2014 persons in New South Wales. Calculation of life expectancy requires mortality rates at all ages.

The authors’ approach to this was complex, involving the derivation of standardized mortality ratios (SMRs) relative to the 1995–1997 Australian General Population, fitting a ‘smooth curve of SMRs across the life-age spectrum’, ‘blending into a flat extra mortality loading at the oldest ages selected using the observed data to confirm the reducing trend in SMRs at advanced ages’. The SMRs they actually used were not provided in the study.

We re-calculated the life expectancies using empirical age-specific mortality rates computed directly from the data in Table 5 of the study. We note that this is appropriate because there was no secular trend over the time period. The rates over the period are thus as relevant today as they were at any point during the period.

As an example, Table 1 here reproduces Middleton’s values for the case of C5–C8ABC injuries. This simple approach, which requires no major assumptions, yields the life table shown here as Table 2. For example, the life expectancy at age 25 is 36.5 additional years, which represents 66% of the Australian general population figure.

Table 1 Data and results derived from Table 5 of Middleton et al.,¹ C5–8ABC group

Table 2 Life table for the C5–C8ABC group, based on empirical mortality rates derived from the data in Table 5 of the Middleton study

Table 3 shows life expectancies expressed as percentages of normal. As may be seen, the percentages reported by the authors in their Table 6 are systematically higher than those computed directly from their own data, especially at older ages. The explanation for this discrepancy awaits clarification.

Table 3 Percentage of normal life expectancy for the C5–8ABC group