- Email: [email protected]
CONSORT
THE LANCET 1 2
McNamee D, Horton R. Lies, damn lies, and reports of RCTs. Lancet 1996; 348: 562. Medical Research Council Lung Cancer. Working Party. Comparison of oral etoposide and standard intravenous multidrug chemotherapy for small-cell lung cancer: a stopped multicentre randomised trial. Lancet 1996; 348: 563–66.
Editor’s reply Black will have cause to complain once more this week. On p 708, we publish the 6-year results of a randomised study to investigate conservation therapy for breast cancer. Again, there is no information on the eligible population. Why? Because this figure is simply not available, according to the authors. As John Hampton pointed out over 15 years ago, “It is essential to know this total, for without it it is impossible to be sure about the applicability of the results of the trial.”1 I share Black’s concern that external validity is a neglected aspect of interpretive criticism.2 However, describing the source population is clearly not always straightforward. Only when the generalisability of the primary outcome measure is seen as an important end point in itself will this aspect of the trial report be improved. The widespread implementation of CONSORT should focus much needed attention on this often overlooked issue. Richard Horton The Lancet, 42 Bedford Square, London WC1B 3SL. UK
1 2
Hampton JR. Presentation and analysis of the results of clinical trials in cardiovascular disease. BMJ 1981; 282: 1371-73 Horton R. The interpretive turn. Lancet 1995; 346: 3.
ratio 2·8, p<0·001), supporting previous findings2 and perhaps reflecting the poor tolerance of old people to trauma. Outbreaks of infectious disease were not reported but there were risks—such as the lack of drinking water because of damage to over 100 wells and the scarcity of food and shelter. Contaminated ponds used for bathing and cattle watering were a worry and it was recommended that they be drained. We examined tubewells and household water stocks by the test papers for coliforms and general bacteria and told the villagers of any contaminated water sources. We also distributed 2000 faucet-equipped plastic water containers together with health education messages. A US study showed that going to the basement, staying away from windows, and covering the body with a blanket prevent injury during tornadoes,3 but housing conditions and, thus, feasible behaviours vary from country to country. Developing countries need to have in place practical and inexpensive measures, using existing resources to minimise the health impact of disasters, and the international community could do more to provide assistance with disaster prevention (eg, weather warning systems). *Osamu Kunii, Takeaki Kunori, Kodo Takahashi, Masaki Kaneda, Nobuo Fuke *International Medical Centre of Japan, 1-21-1 Toyama, Shinjuku, Tokyo 162, Japan; Toyko Medical College, Toyko; Nippon Medical School/Chiba Hokuso Hospital; Chiba; St Marianna University/Toyoko Hospital, Kanagawa; and Teikyo University/Ichihara Hospital, Chiba
1 2 3
Brenner SA, Noji EK. Head and neck injuries from 1990 Illinois tornado. Am J Publ Health 1992; 82: 1296–97. Carter AO, Millson ME, Allen DE. Epidemiologic study of deaths and injuries due to tornadoes. Am J Epidemiol 1989; 130: 1209–18. Duclos PJ, Ing RT. Injuries and risk factors for injuries from the 29 May 1982 tornado Marion, Illinois. Int J Epidemiol 1989; 18: 213–19.
Health impact of 1996 tornado in Bangladesh SIR—On the evening of May 13, 1996, a tornado devastated 90 villages in north-central Bangladesh. 558 people died and over 40 000 were injured. 33 000 houses (70–90% of standing structures in the villages) were destroyed within 20 minutes. On May 17 the Japanese government dispatched the Japan Medical Team for Disaster Relief (JMTDR), which has access to over 500 medical and logistics experts specially trained for overseas disaster relief. In a mobile clinic and a referral hospital, we treated 361 patients (955 patientdays) in 2 weeks. Of the 361 patients, 99% had multiple trauma. The most common injuries (82%) were cuts, 42% of which had skin defects of 10 cm or more; 63% of cuts were deep, reaching muscle. Three-quarters of the wounds were caused by flying corrugated iron sheets used as roofs and walls. Fractures were sustained in 21% of patients, half being open ones. 84% of wounds were infected. The injury sites were the legs (45%), arms (34%), head (34%), and back (20%). As expected,1 head-and-neck injuries were more common in those who died. Referral to another hospital was indicated in 21% of cases but, for financial reasons, few patients could be transferred. There were no systematic warnings of the tornado; in any case 73% of the patients had no radio or television. 94% had not known that disaster was imminent and when the tornado struck 75% of victims were indoors; 12% of those lay on the floor but the rest took no evasive action. Of the 25% who were outdoors, 57% ran back home and 29% jumped into the river; the other 14% took refuge in a concrete building but when that collapsed many deaths and serious injuries resulted. In a sample of 1416 people in households where there had been casualties, the injury and fatality rates were 60% and 6%, respectively. Those over 50 were more likely to die (odds
Vol 348 • September 14, 1996
Travel advice from embassies SIR—One would hope that embassies and consulates would be accurate sources of health information for travellers planning to visit their countries. A decade ago a study of embassies and consulates of developing countries indicated that “only 28% provided complete and correct” travel advice, especially on inoculations and malaria prophylaxis.1 We have repeated the study, using similar methods. 60 embassies and 17 consulates of African, Asian, and South and Central American countries listed in the Ottawa and Toronto official directories were telephoned. Callers posed as travellers inquiring whether “any shots or pills” were recommended. Of the 77 embassies and consulates, 30% provided complete and correct information on yellow fever immunisation and malaria prophylaxis as recommended by the US Centers for Disease Control and Prevention. 36% (18% in the previous study) referred us to other sources of information such as family physicians, travel clinics, and Internet sites. 20 embassies and consulates had a world-wide-web site; four sites provided health advice, all of which was correct. In contrast, 39% did not indicate that any precautions were necessary. Of these, two-thirds were representing countries in areas of significant risk for malaria and one-fifth were endemic regions for yellow fever. Several diplomatic staff falsely assured us that there were no health risks and wished us “bon voyage”. In 38% immunisation recommendations did not accord with those of the CDC. For example, several West African countries neglected to mention that yellow fever vaccination is an official requirement for entry where the disease is endemic. Only 4% indicated the need for the hepatitis A vaccination, when all countries were in zones where the
757
McNamee D, Horton R. Lies, damn lies, and reports of RCTs. Lancet 1996; 348: 562. Medical Research Council Lung Cancer. Working Party. Comparison of oral etoposide and standard intravenous multidrug chemotherapy for small-cell lung cancer: a stopped multicentre randomised trial. Lancet 1996; 348: 563–66.
Editor’s reply Black will have cause to complain once more this week. On p 708, we publish the 6-year results of a randomised study to investigate conservation therapy for breast cancer. Again, there is no information on the eligible population. Why? Because this figure is simply not available, according to the authors. As John Hampton pointed out over 15 years ago, “It is essential to know this total, for without it it is impossible to be sure about the applicability of the results of the trial.”1 I share Black’s concern that external validity is a neglected aspect of interpretive criticism.2 However, describing the source population is clearly not always straightforward. Only when the generalisability of the primary outcome measure is seen as an important end point in itself will this aspect of the trial report be improved. The widespread implementation of CONSORT should focus much needed attention on this often overlooked issue. Richard Horton The Lancet, 42 Bedford Square, London WC1B 3SL. UK
1 2
Hampton JR. Presentation and analysis of the results of clinical trials in cardiovascular disease. BMJ 1981; 282: 1371-73 Horton R. The interpretive turn. Lancet 1995; 346: 3.
ratio 2·8, p<0·001), supporting previous findings2 and perhaps reflecting the poor tolerance of old people to trauma. Outbreaks of infectious disease were not reported but there were risks—such as the lack of drinking water because of damage to over 100 wells and the scarcity of food and shelter. Contaminated ponds used for bathing and cattle watering were a worry and it was recommended that they be drained. We examined tubewells and household water stocks by the test papers for coliforms and general bacteria and told the villagers of any contaminated water sources. We also distributed 2000 faucet-equipped plastic water containers together with health education messages. A US study showed that going to the basement, staying away from windows, and covering the body with a blanket prevent injury during tornadoes,3 but housing conditions and, thus, feasible behaviours vary from country to country. Developing countries need to have in place practical and inexpensive measures, using existing resources to minimise the health impact of disasters, and the international community could do more to provide assistance with disaster prevention (eg, weather warning systems). *Osamu Kunii, Takeaki Kunori, Kodo Takahashi, Masaki Kaneda, Nobuo Fuke *International Medical Centre of Japan, 1-21-1 Toyama, Shinjuku, Tokyo 162, Japan; Toyko Medical College, Toyko; Nippon Medical School/Chiba Hokuso Hospital; Chiba; St Marianna University/Toyoko Hospital, Kanagawa; and Teikyo University/Ichihara Hospital, Chiba
1 2 3
Brenner SA, Noji EK. Head and neck injuries from 1990 Illinois tornado. Am J Publ Health 1992; 82: 1296–97. Carter AO, Millson ME, Allen DE. Epidemiologic study of deaths and injuries due to tornadoes. Am J Epidemiol 1989; 130: 1209–18. Duclos PJ, Ing RT. Injuries and risk factors for injuries from the 29 May 1982 tornado Marion, Illinois. Int J Epidemiol 1989; 18: 213–19.
Health impact of 1996 tornado in Bangladesh SIR—On the evening of May 13, 1996, a tornado devastated 90 villages in north-central Bangladesh. 558 people died and over 40 000 were injured. 33 000 houses (70–90% of standing structures in the villages) were destroyed within 20 minutes. On May 17 the Japanese government dispatched the Japan Medical Team for Disaster Relief (JMTDR), which has access to over 500 medical and logistics experts specially trained for overseas disaster relief. In a mobile clinic and a referral hospital, we treated 361 patients (955 patientdays) in 2 weeks. Of the 361 patients, 99% had multiple trauma. The most common injuries (82%) were cuts, 42% of which had skin defects of 10 cm or more; 63% of cuts were deep, reaching muscle. Three-quarters of the wounds were caused by flying corrugated iron sheets used as roofs and walls. Fractures were sustained in 21% of patients, half being open ones. 84% of wounds were infected. The injury sites were the legs (45%), arms (34%), head (34%), and back (20%). As expected,1 head-and-neck injuries were more common in those who died. Referral to another hospital was indicated in 21% of cases but, for financial reasons, few patients could be transferred. There were no systematic warnings of the tornado; in any case 73% of the patients had no radio or television. 94% had not known that disaster was imminent and when the tornado struck 75% of victims were indoors; 12% of those lay on the floor but the rest took no evasive action. Of the 25% who were outdoors, 57% ran back home and 29% jumped into the river; the other 14% took refuge in a concrete building but when that collapsed many deaths and serious injuries resulted. In a sample of 1416 people in households where there had been casualties, the injury and fatality rates were 60% and 6%, respectively. Those over 50 were more likely to die (odds
Vol 348 • September 14, 1996
Travel advice from embassies SIR—One would hope that embassies and consulates would be accurate sources of health information for travellers planning to visit their countries. A decade ago a study of embassies and consulates of developing countries indicated that “only 28% provided complete and correct” travel advice, especially on inoculations and malaria prophylaxis.1 We have repeated the study, using similar methods. 60 embassies and 17 consulates of African, Asian, and South and Central American countries listed in the Ottawa and Toronto official directories were telephoned. Callers posed as travellers inquiring whether “any shots or pills” were recommended. Of the 77 embassies and consulates, 30% provided complete and correct information on yellow fever immunisation and malaria prophylaxis as recommended by the US Centers for Disease Control and Prevention. 36% (18% in the previous study) referred us to other sources of information such as family physicians, travel clinics, and Internet sites. 20 embassies and consulates had a world-wide-web site; four sites provided health advice, all of which was correct. In contrast, 39% did not indicate that any precautions were necessary. Of these, two-thirds were representing countries in areas of significant risk for malaria and one-fifth were endemic regions for yellow fever. Several diplomatic staff falsely assured us that there were no health risks and wished us “bon voyage”. In 38% immunisation recommendations did not accord with those of the CDC. For example, several West African countries neglected to mention that yellow fever vaccination is an official requirement for entry where the disease is endemic. Only 4% indicated the need for the hepatitis A vaccination, when all countries were in zones where the
757