Survival after SCI

• An international first
• A review of the international literature

An international first

There have been only two studies of survival from SCI conducted in Australia (Yeo et al, 1998; Sneddon & Bedbrook, 1982). Neither was based on a defined population and neither was national in scope. There has been only one SCI survival study in the world that was national in scope, but that was restricted to a veteran population (Samsa, 1993). Australia is one of the few countries that have a national population based register of SCI, providing the opportunity for the first national survival study of a non-veteran community (O'Connor, 1999b). The aims of this study are (1) to determine survival from SCI in a national population, and (2) to help to prevent premature death by identifying the leading causes and some of their risk factors. The ASCIR has been linked with the Australian National Deaths Index (NDI) to determine which members of the SCI population have died, when they died and what caused their death. The NDI is a database, housed at the Australian Institute of Health and Welfare, which contains records of all deaths occurring in Australia since 1980. The data are obtained from the Registrars of Births, Deaths and Marriages in each State and Territory.

The results of the study will be reported in a future issue of the Bulletin. In this issue, a review of the international literature is summarised.

A review of the international literature

The results of the most often cited studies of survival after SCI are summarised in Table 3.

Toward the end of World War II the acute mortality rate from SCI reportedly declined to approximately 20% (Poer, 1946; Barber & Cross, 1952). Whiteneck et al. (1992) contrast the median survival for those injured during the 1940's (26 years) with those injured in the 1960's (33 years). These changes were reported to be largely due to the availability of penicillin, sulfa drugs and whole blood (DeVivo et al., 1987).

Despite the substantial number of studies of survival from SCI since that time, it is difficult to determine the extent of further improvement due to substantial differences in the study methods employed. Referring to Table 3, the following differences are evident:

• Different injury and follow-up periods. Follow-up period is not specified in some articles. There seems to be no uniformity as to the time interval that constitutes the acute phase following injury. Studies focus on deaths after one day, two weeks, three months, one year or eighteen months.
• Different populations ie. different countries, different selection criteria, veterans vs. general population, restriction to survival periods that vary from study to study, and losses to follow-up.
• Some studies were based on small case series.
• Different survival parameters, especially variation in the cumulative survival period reported.
• Results reported for different groups and time periods, confidence intervals not often reported, and there was generally a failure to stratify the results by important prognostic factors such as age and level and completeness of injury.
• Relative survival was not always provided, making it difficult to compare the results of the different countries which have different survival rates in the general population.
• Few of the studies were population based and only one was national in scope, but restricted to veterans.

Given these differences it is not surprising that interpretations of the literature vary.

DeVivo et al. (1992b) found an increase, not statistically significant, in the Year 2 survival percentage from 90.0% in 1973-77 to 94.1% in 1984-86. DeVivo et al. (1992a) found an increase in cumulative seven year survival from 86.7% in 1973-80 to 89.2% in 1973-84. One of the most recent studies (Hartkopp et al., 1997a) found a 10% increase in the 10 year survival probability for men and 21% increase for women between cases injured in 1953-71 compared with 1972-90, although a corrigendum (Hartkopp et al., 1997b) raised doubt over these figures. Comparing their results with those of Geisler (1983) a decade earlier, McColl et al. (1997) found an increase in total life expectancy of nearly five years, a result which they suggest is similar to that found by Whiteneck et al. (1992). However, for the reasons already discussed, it is difficult to be sure of the extent of improvement in survival.

The studies reported in Table 3 suggest that the 20-25 year survival of SCI cases could range from about 70% to 90% of that of the general population whilst the 40 year survival could be about 70% (based on the experience of veterans). However, this assessment is based on the small number of studies that have reported relative survival or have presented data that enables it to be estimated.

There are a number of factors that have been consistently shown to be predictors of mortality for the SCI population. Age at injury and year of injury are two such factors. Those injured at younger ages and those injured more recently experience better survival outcomes (McColl et al., 1997; Whiteneck et al., 1992; Samsa et al., 1993; Hackler, 1977; Kiwerski, 1993). Neurological level and completeness of injury have also been found to be important predictors of mortality (McColl et al., 1997; Whiteneck et al., 1992; Hackler, 1977; Kiwerski, 1993). Early mortality for those aged 50 or more years with complete cervical cord injury is very high. Alander et al. (1994) found a 60% within the first four months. The intermediate-term outcome for this group was also poor, with only 13% surviving the first year (Alander et al., 1997). Those with incomplete injuries had a better outcome in the intermediate-term, with 93% surviving the first year, but only 50% survived to 5.5 years (Alander et al., 1997).

Changes in medical practice from the early 1970s have brought about a significant change in the causes of death. Prior to that time renal failure was the primary cause of death (Breithaupt et al., 1961; Freed et al., 1966; Jousse et al., 1968). Substantial decreases in respiratory and renal diseases have occurred since that time (Kraus et al., 1979; Geisler et al., 1983; DeVivo et al., 1989). Whiteneck et al. (1992) observe that whilst genitourinary disorders accounted for 43% of deaths occurring in the 1940s and 1950s, they accounted for only 10% of deaths occurring in the 1980s and 1990s. Geisler (1983) speculated that advances in the fields of antibiotics, renal dialysis and in medical and surgical expertise contributed to the decline in these diseases.

As the survival of the SCI population approaches that of the general population, the causes of death also appear to be approaching those of the general population (Samsa, 1993; Whiteneck et al., 1992). However, the illnesses that cause death occur at younger ages in the SCI population (Whiteneck et al. (1992). Also many of the cause specific death rates for SCI remain substantially above the normal population, particularly for infections, primarily septicaemia and pneumonia (Hartkopp et al., 1997; DeVivo et al., 1993; Samsa et al., 1993), but also for pulmonary emboli and suicide (DeVivo et al., 1991; Charlifue & Gerhart, 1991; Hartkopp et al., 1997a, 1997b). Hartkopp et al. (1997a) found that there was a high percentage of deaths due to lung disease, especially pneumonia, amongst those with functionally complete tetraplegia (76%).

The problems of an aging SCI population have only recently begun to be studied and addressed. Only a few studies of long-term survival (greater than 20 years) have been published. McColl et al. (1997) questions whether as survival increases, mortality continues to be compressed into the last few years of life or whether morbidity expands to fill the increased survival period. Their study of health expectancy favours the more optimistic view. Whiteneck et al. (1992) emphasise the difficulty in separating the effects of age from the effects of chronic disability in individuals who are experiencing both simultaneously. The inherent confounding of these factors makes analysis complicated.

These are some of the issues that the Australian survival study will address.