THE EFFECT OF CLIMaTE UPOK HEALtH AjS'D DISEASF. 4o8 



represent the true death-rate on the hospital population, aa a certain 

 proportion die after discharge from hospital. 



In the special case of notifiable diseases we have, on the one hand, 

 the total cases notified, and, on the other, the total deaths from each 

 disease. One may assume that the fatal cases are notified, and that the 

 total mortality on population is approximately correct : but it does not 

 appear that the attack-rate calculated from the cases notified can always 

 be accepted as accurate, or that two attack-rates in different localities 

 are comparable, as there are considerable differences in different areas in 

 the carrying out of the process of notification, especially abroad. 



Practically, where accurate methods of incidence are required, and of 

 mortality in acute disease, we have only the naval and military statistics 

 to depend on. The records from prisons and gaols are more difficult of 

 classification ; certainly the total sickness of the population is dealt with, 

 but the periods of ' exposure to risk ' vary so much as to complicate 

 matters. The records from the Indian (and from some Colonial) gaols 

 are, however, extremely valuable in spite of this complication. 



In all these cases the number of fresh attacks, as distinct from 

 relapses, in such diseases as, e.g., malarial fever and dj«entery, should be 

 laoted ; special records are usually necessary to elucidate this point, which 

 is most important. 



(ii) Where deaths only are recorded. 



These are available, with varying degrees of accuracy, in all civilised 

 countries. They form in many cases the only available records which 

 relate to the whole, or, indeed, to any part of the population. They are, 

 or can be made, available by age groups. Unfortunately, they have not 

 all the same degree of accuracy ; in the younger or less civilised countries 

 the case often ends fatally without the cognisance of a medical man, and 

 a medical certificate is not always obligatory, as indeed it often could 

 not be obtained. Hence the deaths cannot be properly enumerated for 

 statistical purposes ; it is often not possible even to guess from the entry 

 in the register what the cause of death may have been. Again, even 

 where a medical certificate has been furnished the cause of death is too 

 often described in vague or even misleading terms. The inquirer must 

 often examine the register himself, and as conscientiously as possible 

 eliminate and reconstruct. This reconstruction is, of course, a dangerous 

 process ; it can only give possibilities, or at most probabilities. But in 

 some instances the true meaning of the death certificate is perfectly 

 obvious, especially where the inquirer has local knowledge. 



(rf) The compilation of statistical tables of disease. 

 The useful periods are : — 



(i) The week. Probably only useful in epidemic disease, or where a 

 disease has a distinct seasonal variation of short period. Except for the 

 United Kingdom there are no regular meteorological tables with which 

 to compare these weekly results ; the observer must then re-compile either 

 (rt) the daily meteorological data, to suit his weekly table, or {h) compile 

 a monthly disease table from his daily records, as the weekly tables will 

 not of themselves give monthly tables. The former is certainly the more 

 useful process in the conditions in which a weekly disease record is 

 wanted ; the second process defeats the purpose of the weekly record ; 

 but in the end both weekly and monthly tables will usually be required. 



