Each hospital admission is assigned a main diagnosis of admission, and each death is assigned an underlying cause of death (secondary diagnosis codes and causes can also be given if there are other influential medical factors or causes for the admission or death). The system used to code these diagnoses and causes is the International Classification of Diseases. This not only consists of codes for all the known medical conditions and diseases, but also other reasons, causes or factors for the admission or death. These other codes include routine codes for hospital care, such as those related to screening, diagnosis or maternity, and other external factors such as traffic accidents, injuries, falls, poisoning, suicide, etc. It is possible that not all relevant secondary diagnoses codes will be coded, and included on the admission or mortality data as it will depend on what doctors write and record in their medical notes or on the death certificate.
Each of the ICD codes have been assigned a ‘smoking attributable fraction’ (SAF) which has been determined through research and analysis of hospital admission and death data nationally. It is not based on ‘real’ data, and only represents a modelled estimate of the numbers and rates of smoking-related admissions and deaths. The assigned SAF values ranges from zero to one. The SAF are based on quite rigorous research, so are probably quite accurate. In the case of SAF values, they are assigned to the ICD 10 codes based on the primary diagnosis of the admission or the underlying cause of the death (in contrast to Alcohol Attributable Fractions which do use secondary diagnosis codes for admission or death, but for alcohol it is more apparent when a proportion of some external causes will have been caused by alcohol such as road traffic accidents, poisoning or falls, etc). The same methodology and calculations apply as to how the SAFs are applied to admissions or deaths. Different SAF values are assigned on the basis of the diagnosis codes for different age groups, and the value of the SAF often differ between males and females. Many conditions will be assigned a value of zero as there is no current evidence that they are related to smoking at all, but many respiratory diseases and cancers may have a value between zero and one where there is evidence that smoking is (sometimes) involved or is a contributing factor to the disease.
Estimates of the number and rate of admissions and deaths that are attributable to smoking are given on Public Health England’s Fingertips. The sum of all the SAF values for all hospital admissions (based on primary diagnosis code of admission) and for all deaths (based on underlying cause of death) are summed to give an estimate of the total number of estimated admissions and deaths attributable to smoking. Further information and examples can be found in the glossary in relation to Alcohol Attributable Fractions although the calculations are more complex for alcohol with secondary diagnoses codes having some influence), but the SAF combines information in relation to smoking prevalence (which the Alcohol Attributable Fractions do not).
Royal College of Physicians. Hiding in plain sight. Treating tobacco dependency in the NHS. A report by the Tobacco Advisory Group of the Royal College of Physicians. London: RCP, 2018.
Also see International Classification of Diseases and Alcohol Attributable Fractions.
Also see Directly Standardised Rate.