Wednesday 6 May 2009

Mortality and Morbidity Risk in the L'Aquila Earthquake and Some Lessons to be Learned



The L'Aquila earthquake (Mw=6.3) occurred at 03.32 hrs local time when most people were sleeping. Analyses of world-wide patterns of casualties suggest that between 50 and 90 per cent of deaths in earthquakes occur between midnight and 6 a.m. (as seismic casualty data are notoriously irregular, the difference depends on the period to which records pertain--Alexander 1996, Jones et al. 1990). Studies in central America and Turkey highlight the importance of vernacular housing as a source of risk in nocturnal earthquakes, or, indeed, whenever people are likely to be at home (Glass et al. 1977, Angus 1997, Rodriguez 2005). That is equally true in Italy (De Bruycker et al. 1983, 1985) where the only buildings that are more vulnerable to collapse (and may on occasion be fully occupied) are ecclesiastical ones. Some of them are very large, extremely old, poorly maintained and lacking in seismic retrofit provisions.

Examination of patterns of damage in the L'Aquila earthquake suggests that it may be possible to create model damage scenarios to help examine the question of earthquake survivability. Two examples follow.

Model URM vernacular dwelling. A typical unreinforced masonry (URM) single family vernacular dwelling in a village (such as Onna) or small town of Abruzzo Region might have the following characteristics:-
* two or three storeys with an independent entrance but bounded laterally by other dwellings
* rubble masonry vertical load-bearing walls 30-40 cm thick consisting of angular limestone fragments bound together with soft lime mortar and cement rendered or covered with stucco
* hard spots caused by localised repairs, usually about 1-3 sq. metres in size
* weak zones located primarily between apertures, at roof level and at corners, or connected with utility channels and chimney recesses in walls
* a heavy roof consisting of a concrete base or assemblage of concrete, steel joists and hollow terracotta tiles overlain with asphalt sheeting and terracotta pantiles; alternatively one laid upon longitudinal wooden beams of 20-20 cm section and spacing approximately 1 m
* chimneys may consist of precast cement segments that detach and collapse during the shaking.

The ancient practice of using courses of tiles in rubble walls, which was started by the Romans and continued until the early 20th century, could be seen in a minority of buildings in L'Aquila province. It contributed to their cohesion but not to the extent of providing full anti-seismic protection.

As it weakened heavy masonry walls that lacked basic structural integrity, the practice of carving channels in walls for plumbing and electrical lines (chasements) led to many failures during the 6 April 2009 earthquake. Recesses and channels for chimneys had a similar effect.

Many failures in URM buildings were connected with mixed construction, as where rubble masonry in the original building was augmented by brick, cement block or concrete alterations (or even all three). Differing stiffness, compressibility and weight of these components would tend to complicate a building's reaction to seismic stresses.

Model RC vernacular dwelling. A typical reinforced concrete (RC) vernacular dwelling in an Abruzzo town or in L'Aquila city might be characterised as follows:-
* a three-to-five storey multiple-family condominium with a communal entrance and communal stairs
* use of smooth reinforcing bars (until the 1970s); over-economical usage and poor positioning of stirrups, poor design or construction of joints
* a heavy concrete roof with tile overlay
* hollow-brick infill wall panels that are poorly tied to the frame and may fall out or inwards
* thin, hollow-brick internal partition walls.

As a result of racking of the frame, infill wall panels tended to detach from the frame and fall in or out, perhaps fragmented by X-shaped cracking. Similarly, partition walls fractured and collapsed inside the buildings. Racking also causes pounding, fracturing and torsion at structural nodes. In some cases, the stairs detached from supports and collapsed. Finally, there were instances of heavy damage to plaster, ceilings and fixtures and overturning of furniture.

In L'Aquila there were many examples of incipient (or actual) mid-floor failure in multi-storey RC dwellings. This is indicative of inadequate stiffness with inertial effects above coupled with heavy displacement below. The latter may have been affected by seismic wave amplification in alluvial or lacustrine sediments or topographic amplification on convex hillslopes. In many cases this did not lead to collapse of the building but internal damage (i.e. to partition and infill walls) was very substantial.

With regard to both sorts of dwelling the modern practice of laying terracotta tiles on asphalt sheeting, such that the only things that secure them are weight and interlocking friction, led to the displacement of large numbers of tiles into the street. The lightest form of roofing tile used in Italy (20 x 36 cm) weighs about 1 kg, which amounts to 15 kg/m2. Curved pantiles are at least 60-100 per cent heavier than this. It is thus easy for heavy agglomerations of tiles to cascade over the edge of roofs into the street and to take cornices, balcony stonework and façade details with them.

Buildings that were not damaged to the point of partial or total collapse showed surprisingly little breakage of window glass. In other earthquakes this has been a factor in injuring people who rushed outside without adequate footwear. Likewise, collapse of light fittings was not widespread enough to create a significant glass splinter hazard.

Damage and the potential to improve instantaneous self-protective reaction

Given the complexity of failure patterns in vernacular housing it is reasonable to suppose that there is no single self-protective behaviour that would be appropriate under all scenarios for damage. Despite the controversy over the predictability of the L'Aquila earthquake,[1] it remained unexpected and very few people were prepared for it when it happened.

The obstacles to immediate and short-term earthquake preparedness fall into six categories:-
* experience: people may lack experience or have had no direct contact with the problem
* adaptability: people may fail to adapt or even perceive the need to adapt to the seismic threat
* perception may be insufficient to enable a person to understand the problem well enough to be motivated to act
* social: failure to communicate, associate and learn
* economic: failure or inability to accumulate money and invest in protection
* organisational: lack of social structure and incentive to act.

Factors that increase the risk of injury in the case of rapid exit from a building include the following:-
* battering by adjacent structures
* collapse of URM walls, as coherent slabs or in fragments
* detachment of roofs
* detachment and collapse of pinnacles, balustrades and chimneys
* demolition by falling masonry of balconies and façade details that jut out
* separation of URM walls from roofs, with collapse of cornices and upper masonry
* ejection of infill walls in RC buildings
* detachment and collapse of corners in URM buildings
* detachment and collapse of stairs
* racking distortion of apertures.

On the other hand, these are some of the factors that increase the risk of injury in the case of deciding to remain inside a building:-
* battering by detached horizontal members (wooden roof beams and steel floor joists)
* torsion, distortion and shattering of nodes in RC buildings
* detachment of roofs
* bulging and reticular cracking of walls, with detachment of rendering and stucco and eventual collapse of the structure
* X-shaped, diagonal or reticulated cracking in the weak zones between apertures
* implosion of infill walls in RC buildings and collapse of internal partition walls
* damage to ceilings and internal fittings and overturning of furniture.

In heavily damaged buildings in L'Aquila there was little indication that the "triangle of life" would have helped to save people from crush injuries or being buried by dust and rubble. Neither would sheltering under tables or desks.

The "triangle of life" has been vigorously promoted by the American Rescue Team (see www.amerrescue.org) but equally vigorously contested by other protagonists (Lopes 2004). It involves sheltering next to large, robust objects that block the collapse of beams and slabs and leave a triangular cavity in which a person may shelter (relatively) unscathed. In general, complete collapse of a frame building may leave some void spaces, perhaps 10-15 per cent of the resulting mound of rubble, but they can easily fill with cement, gypsum or mortar dust and fragments. Examination of the partial and total collapse of buildings in L'Aquila city and Onna suggested that the "triangle of life" approach would have been ineffective as few such cavities were present.

There is some--albeit circumstantial--evidence that when buildings were being heavily damaged the best spontaneous action would have been to retreat further inside. Running into the street would put people significantly at risk from falling masonry or the collapse of stairways. In any case, rapid egress was made difficult by doors that jammed as a result of racking distortion.

In consideration of the types and levels of damage caused in the L'Aquila earthquake, risk of death or injury can be related to damage level on the following five-point scale:-

1. Damage level: minimal indoor damage to walls, fixtures and fittings.
Personal risk: for most people, prudent behaviour ensures freedom from injury.
2. Damage level: significant damage to structure and fittings.
Personal risk: risk of moderate injury but no significant risk of death.

3. Damage level: pervasive damage and collapse of architectural details.
Personal risk: significant risk of serious injury but low risk of death.

4. Damage level: major damage and limited partial collapse.
Personal risk: strong risk of serious injury and significant risk of death.

5. Damage level: collapse of more than 50 per cent of the structure.
Personal risk: limited probability of survival.

Independently of any question of making buildings safer by retrofitting them, it would be possible to create a strategy to survive earthquakes while at home--at least under ideal circumstances of perception and commitment of householders. This would involve making an educated guess about the probable seismic behaviour of a vernacular dwelling and planning to react accordingly. The following steps are proposed:-

* Identify and avoid the riskiest forms of behaviour, such as running blindly out of the house.
* Develop criteria to identify the safest place in the house--i.e. the most robust place with the least risk of collapse--in the light of the following considerations:
- potential for detachment and displacement of roof tiles or the entire roof
- stability of cornices and external balusters
- degree of support of staircases
- possibility of battering interference with adjacent buildings that are different in size, shape and construction and thus have different fundamental periods
- heterogeneity of materials and potential for interference or complex behaviour.
* Create an egress procedure, considering the difficulties of exiting a building in an environment characterised by high levels of damage and precariousness. The procedure should identify the nearest safe refuge and assembly area.
* Identify the most dangerous places in the house and plan to withdraw from them.
* Create a mutual support network of relatives, friends and neighbours.
* Assemble a cache of small-scale emergency equipment and materials (torch, radio, hard hat, water sterilisation pills, etc).
* Instruct and train family members and ensure that drills are practiced.

The presence of an elementary school in the middle of the urban area in Onna that was of new construction and which resisted the earthquake without damage is an indication of the importance of such buildings as the potential location of command posts, points of refuge for the population and reception centres for people who cannot return home. Ideally, each neighbourhood or village should have such a building. It should specifically be designated as multi-function and should be equipped accordingly.

Scenarios for earthquakes at other times of day

Since pioneering work in Chile in 1960 (Lomnitz 1970) it has been well-known that aggregate patterns of human behaviour can have a very substantial impact on the totals and patterns of earthquake injury epidemiology. In this respect it is interesting to speculate on what the situation would have been if the L'Aquila earthquake had occurred at another time of day (and on a holiday or working day).

In the L'Aquila earthquake there was an overall death/injury ratio of 0.20 (305 deaths--plus two related heart attack fatalities--and about 1500 recorded injuries)[2], which is relatively low for medium-to-large earthquakes (0.33 has been hypothesised--PAHO 1981). The case fatality rates of 0.17 overall and 0.60 for serious and critical (hospitalised) injuries are low in the first case and high in the second, as the ratio of serious to all injuries was only 0.13, which is somewhat small by comparison with similar earthquakes elsewhere (commonly it might be 0.15-0.25).

Would it have been much different if the earthquake had occurred at another time of day or not on a Sunday or holiday?

Damage to religious buildings was serious enough that if the tremors had occurred during Sunday mass (as happened at Lisbon in 1755--Chester, 2001--and in Irpinia-Basilicata, southern Italy, in 1980--De Bruycker et al. 1985) death tolls among congregations would inevitably have been high. The spontaneous collapse of the vaulting of the Upper Basilica in Assisi after the 1997 Umbria-Marche earthquake swarm crushed four people to death and provided a clear illustration of what could happen to congregations. Moreover, 81 died in the collapse of the church in Balvano, Potenza Province in 1980. Like many churches in the Province of L'Aquila it lacked any significant resistance to seismic acceleration.

Damage to public buildings was substantial but, with the exception of the Prefecture (Palazzo del Governo), which largely collapsed, it appears to have been less than that inflicted upon vernacular housing. However, cornice collapse and shedding of rubble and roofing material into streets could have caused a significant number of fatalities and injuries (including people in cars) if the streets had been busily occupied rather than deserted, especially in the commercial cores of the city and neighbouring towns. This alone might have led to an even greater death toll.

Significant non-structural damage occurred to the L'Aquila city bus station, a steel-framed building with brick cladding. However, it is only a one-storey building and if it had been full of people there would probably have been significant injuries but few or no deaths.

Serious damage occurred to commercial and industrial premises, but in these injury tolls would probably have been limited by low density of occupancy. However, at the main hospital in L'Aquila there was significant potential for a greater number of injuries if the earthquake had occurred during the day when many more people would have been using this complex of buildings. As the damage was limited to cladding, ceiling fixtures and walls, no one died in the hospital and that would probably still have been the case if it had been more fully occupied. Nevertheless, injuries might have been concentrated around the main staircase, where damage was more substantial as a result of interference between the two structural masses of the building. Had the earthquake been stronger or more prolonged, the stairs might have collapsed and at certain times of day they could easily have been full of people trying to escape the tremors.

Distribution of fatalities in the L'Aquila earthquake

The economic viability of human settlements in Abruzzo is often related to their demographic growth or decline. Generally, the smaller, more rural or isolated settlements lose population to the larger ones where economic opportunity is greater. In Abruzzo Region a total of 81 municipalities were affected by the earthquake, and 49 of them were inserted into the Prime Ministerial Decree regarding damage of MCS intensities VI-IX.[3] Although there is considerable statistical variation (relating mainly to employment opportunities in the L'Aquila area and close to the Adriatic Sea coast and its access roads), the break-even point that divides decline from growth (measured on the basis of changes over the period 2001-7) is a population of about 1,500, which is the same as it was at the time of the last significant earthquake in the region (Alexander 1986). It is interesting to note that the eight municipalities in which fatalities occurred are all growing, on average by a healthy 3.7 per cent per decade. If deaths can be connected with building collapse in areas of poor quality housing, demographic stagnation is certainly not a factor.

The distribution of the 305 deaths involves a relatively circumscribed area 24 x 11 km in size. The density of population plays some role, as does the geotechnical and geomorphological setting, especially regarding soft sediments and piedmont location.

In considering the age and gender pattern of fatalities, it is of note that they are dominated by the 20-29 and over 70s age groups. The prevalence of mortality among old people is a common feature of major earthquakes (Liang et al. 2001), as they are less mobile, less perceptive and more frail than younger people, and they may live, as pensioners, in poorer quality housing. Moreover, the preponderance of female over male victims among the over-70s probably reflects nothing more than the greater longevity of women. However, the peak in the 20-29 age group is interesting and corresponds to findings from the Kobe earthquake of January 1995 (Osaki and Minowa 2001). This group is highly active but may lack experience of earthquakes and have little idea about what to do during them. Finally, there is a gender bias in the data that cannot be explained purely by the longevity of women. On average 43 men died to every 50 women. If the over 70s are excluded, the figure remains 47.5 men to 50 women. It begs further investigation.

References

Alexander, D.E. 1986. Disaster preparedness and the 1984 earthquakes in central Italy. Working Paper 55, Natural Hazards Center, Boulder, Colorado, 90 pp.

Alexander, D.E. 1996. The health effects of earthquakes in the mid-1990s. Disasters 20(3): 231-247.

Angus, D.C. 1997. Epidemiologic assessment of mortality, building collapse pattern, and medical response after the 1992 earthquake in Turkey. Prehospital and Disaster Medicine 12: 222-234.

Chester, D. K. 2001. The 1755 Lisbon earthquake. Progress in Physical Geography 25(3): 363-383.

De Bruycker, M., D. Greco, I. Annino, M.A. Stazi, N. De Ruggiero, M. Triassi, Y.P. De Kettenis and M.F. Lechat 1983. The 1980 earthquake in southern Italy: rescue of trapped victims and mortality. Bulletin of the World Health Organization 61(6): 1021-1025.

De Bruycker, M., Greco, D. and Lechat, M.F., 1985. The 1980 earthquake in southern Italy: mortality and morbidity. International Journal of Epidemiology 14: 113-117.

Glass, R.I., Urrutia, J.J., Sibony, S., Smith, H., Garcia, B. and Rizzo, L., 1977. Earthquake injuries related to housing in a Guatemalan village. Science 197: 638-643.

Jones, N.P., E.K. Noji, F. Krimgold and G.S. Smith 1990. Considerations in the epidemiology of earthquake injuries. Earthquake Spectra 6: 507-528.

Liang, N.J., Y-T. Shih, F-Y. Shih, H-M. Wu, H-J. Wang, S-F.Shi, M-Y. Liu and B.B. Wang 2001. Disaster epidemiology and medical response in the Chi-Chi earthquake in Taiwan. Annals of Emergency Medicine 38(5): 549-555.

Lomnitz, C. 1970. Casualties and behaviour of populations during earthquakes. Bulletin of Seismological Society of America 60: 1309-1313.

Lopes, R. 2004. American Red Cross response to 'Triangle of Life' by Doug Copp. http://www.bpaonline.org/Emergencyprep/arc-on-doug-copp.html

Osaki, Y. and M. Minowa 2001. Factors associated with earthquake deaths in the Great Hanshin-Awaji Earthquake, 1995. American Journal of Epidemiology 153(2): 153-156.

PAHO 1981. A Guide to Emergency Health Management After Natural Disasters. Pan American Health Organization, Washington, D.C.

Rodriguez, M.E. 2005. Evaluation and design of masonry dwellings in seismic zones. Earthquake Spectra 21(2): 465-492.

[1] See "Earthquake at L'Aquila, central Italy", http://www.emergency-planning.blogspot.com
[2] A complete list of victims has been published and repeatedly updated by the newspaper Il Centro, see: http//
racconta.kataweb.it/terremotoabruzzo/index.php?sorting=morto_frazione,morto_comune,cognome&cerca=cerca
[3] DPCM no.3 of 16-4-2009, " Individuazione dei comuni danneggiati dagli eventi sismici che hanno colpito la provincia dell'Aquila ed altri comuni della regione Abruzzo il giorno 6 aprile 2009." Presidenza del Consiglio dei Ministri, Rome.