Health Heart
Community Health Profiles

Medical

19. DEATHS FROM SMOKING INDICATOR

TABLE 1 – INDICATOR DESCRIPTION

Information component Page 4 Spine Chart – Indicator 19
Subject category / domain(s) How long we live and what we die of
Indicator name (* Indicator title in health profile) Smoking attributable mortality (“Deaths from smoking”)
PHO with lead responsibility ERPHO
Date of PHO dataset creation 20th December 2006
Indicator definition Deaths attributable to smoking, directly age standardised rate, 35 years +, 2003-5, persons.
Geography England, GOR, Local Authority: Counties, County Districts, Metropolitan County Districts, Unitary Authorities, London Boroughs
Timeliness Annual updates Available from erpho
Rationale:What this indicator purports to measure Rate of deaths that can be attributed to smoking in persons over 35
Rationale:Public Health Importance Smoking still accounts for between 1 in 6 and 1 in 10 of all deaths in England, and accounts for about half of the inequality in death rates between spearhead and non-spearhead areas. It remains the biggest single cause of preventable mortality and morbidity in the world.
Rationale: Purpose behind the inclusion of the indicator To encourage smoking preventionTo focus action on tackling smoking related disease and to help prioritise actions to tackle health inequalitiesTo promote better measurement of smoking prevalenceSmoking related deaths is a powerful proxy measure of overall health and predictor of health care demand.
Rationale:Policy relevance Smoking KillsChoosing HealthSmokefree England
Interpretation: What a high / low level of indicator value means High smoking attributable death rates are indicative of poor population health and high smoking rates. A zero rate is unachievable but lower rates than the best in England are seen in California and Scandinavian countries.
Interpretation: Potential for error due to type of measurement method The method relies on the use of estimates of the contribution of smoking to a range of causes of death derived from the American Cancer Prevention Society II study. It presumes a degree of generalisability of these estimates, and does not take into account any degree of uncertainty so the estimates will be over precise.
Interpretation: Potential for error due to bias and confounding We used the England smoking prevalence rates because the model required not just estimates of current smoking but ex-smoking and non-smoking as well. These figures are not available at local authority level. The method will tend to overestimate smoking related deaths in low prevalence areas and underestimate in high prevalence areas.
Confidence Intervals: Definition and purpose A confidence interval is a range of values that is normally used to describe the uncertainty around a point estimate of a quantity, for example, a mortality rate. This uncertainty arises as factors influencing the indicator are subject to chance occurrences that are inherent in the world around us. These occurrences result in random fluctuations in the indicator value between different areas and time periods. In the case of indicators based on a sample of the population, uncertainty also arises from random differences between the sample and the population itself.The stated value should therefore be considered as only an estimate of the true or ‘underlying’ value. Confidence intervals quantify the uncertainty in this estimate and, generally speaking, describe how much different the point estimate could have been if the underlying conditions stayed the same, but chance had led to a different set of data. The wider is the confidence interval the greater is the uncertainty in the estimate.Confidence intervals are given with a stated probability level. In Health Profiles 2007 this is 95%, and so we say that there is a 95% probability that the interval covers the true value. The use of 95% is arbitrary but is conventional practice in medicine and public health. The confidence intervals have also been used to make comparisons against the national value. For this purpose the national value has been treated as an exact reference value rather than as an estimate and, under these conditions, the interval can be used to test whether the value is statistically significantly different to the national. If the interval includes the national value, the difference is not statistically significant and the value is shown on the health summary chart with a white symbol. If the interval does not include the national value, the difference is statistically significant and the value is shown on the health summary chart with a red or amber symbol depending on whether it is worse or better than the national value respectively.

TABLE 2 – INDICATOR SPECIFICATION

Indicator definition: Variable Deaths attributable to smoking
Indicator definition: Statistic Directly age-standardised rate
Indicator definition: Gender Persons
Indicator definition: age group 35+
Indicator definition: period 2003-5
Indicator definition: scale Per 100,000  European Standard population
Geography: geographies available for this indicator from other providers SHAs, new PCOAvailable from erpho on request
Dimensions of inequality: subgroup analyses of this dataset available from other providers Data can be calculated at MSOA level, and aggregated into deprivation quintiles to calculate within area inequalities. See for examplehttp://www.erpho.org.uk/topics/health_inequalities/profilesmap.aspx
Data extraction: Source ONS for mortality data and population estimatesHealth Survey for England for prevalence of smoking, ex-smoking and non-smokingSAMMEC website (for relative risks)
Data extraction: source URL http://apps.nccd.cdc.gov/sammec/ONS mortality extracts are supplied to PHOs under a non-disclosure agreement and the provisions of the Health Act 2000.Health Survey for Englandhttp://www.dh.gov.uk/en/PublicationsAndStatistics/PublishedSurvey/HealthSurveyForEngland
Data extraction: date December 2006
Numerator: definition Smoking attributable mortality Deaths recorded as having the following ICD codes as underlying cause of death, registered in respective calendar years:

Disease Category ICD10
Malignant Neoplasms
Lip, Oral Cavity, Pharynx C00–C14
Oesophagus C15
Stomach C16
Pancreas C25
Larynx C32
Trachea, Lung, Bronchus C33–C34
Cervix Uteri C53
Kidney and Renal Pelvis C64–C65
Urinary Bladder C67
Acute Myeloid Leukemia C92.0
Cardiovascular Diseases
Ischemic Heart Disease I20–I25
Other Heart Disease I00–I09, I26–I51
Cerebrovascular Disease I60–I69
Atherosclerosis I70
Aortic Aneurysm I71
Other Arterial Disease I72–I78
Respiratory Diseases
Pneumonia, Influenza J10–J18
Bronchitis, Emphysema J40–J42, J43
Chronic Airway Obstruction J44
Numerator: source Death extracts from ONS:ONS PHO death extract 2004ONS PHO death extract 2005ONS PHO death extract 2006
Denominator: definition Mid year local authority population estimates
Denominator: source ONS Mid year population estimates 2003: Published 9th September 2004ONS Mid year population estimates 2004: Published 20th December 2005ONS Mid year population estimates 2005: Published 24th August 2006
Data quality: Accuracy and completeness The accuracy of these estimates is contingent on the underlying accuracy of the three components:

  1. Mortality data – because this is cause specific there may be coding variation between places although for the key contributors e.g. cancer deaths this is less likely.
  2. The relative risks are unpublished and only made available through the SAMMEC website – it is difficult to comment on their validity although they update the previous estimates using in prior estimation of smoking deaths e.g. the Smoking epidemic which used risk estimates from ACPSI.
  3. Smoking prevalence estimates are based on national survey data and are likely to accurate but imprecise.

TABLE 3 – INDICATOR TECHNICAL METHODS

Numerator: extraction The following SQL string was used to extract the relevant count data from the mortality database CREATE TABLE P1578_HPSAMMEC_LA_ICD10 (RegPeriod varchar(8),  LACode varchar(4),  LAName varchar(30),  Sex varchar(1),  AgeGroup varchar(5),  SAFAgeGrp varchar(5),  ICDCode varchar(4),  Deaths decimal(8,3))INSERT INTO P1578_HPSAMMEC_LA_ICD10  (RegPeriod,   LACode,  LAName,  Sex,  AgeGroup,  ICDCode,  Deaths)SELECT  ‘2003-05’,  [LA Code now],  [LA name],  Sex,  [FiveYearTo85],  SUBSTRING([Und COD (non-neo)],1,3),  COUNT(*)FROM M_DEATHS_96_05  LEFT JOIN A_DeathAgeBands ON M_DEATHS_96_05.Age = A_DeathAgeBands.AgeInUnits AND M_DEATHS_96_05.[Age Unit] = A_DeathAgeBands.Units  LEFT JOIN HES.dbo.A_LACounty ON M_DEATHS_96_05.ResCty + M_DEATHS_96_05.ResLAUA = HES.dbo.A_LACounty.[LA Code]   LEFT JOIN HES.dbo.A_NationalPostcodes_May06 ON M_DEATHS_96_05.Pcode = HES.dbo.A_NationalPostcodes_May06.[8digit postcode]WHERE YEAR([Reg Date]) > 2002   AND YEAR([Reg Date]) < 2006   AND [LA Code now] <> ‘NULL’   AND [LA Code now] <> ’15UH’   AND ([FiveYearTo85] = ’35-39′ OR [FiveYearTo85] = ’40-44′ OR [FiveYearTo85] = ’45-49′ OR [FiveYearTo85] = ’50-54′     OR [FiveYearTo85] = ’55-59′ OR [FiveYearTo85] = ’60-64′ OR [FiveYearTo85] = ’65-69′ OR [FiveYearTo85] = ’70-74′     OR [FiveYearTo85] = ’75-79′ OR [FiveYearTo85] = ’80-84′ OR [FiveYearTo85] = ’85+’)  AND (SUBSTRING([Und COD (non-neo)],1,3) = ‘C00’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C01’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C02’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C03’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C04’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C05’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C06’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C07’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C08’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C09’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C10’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C11’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C12’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C13’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C14’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C15’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C16’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C25’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C32’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C33’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C34’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C53’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C64’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C65’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘C67’ OR [Und COD (non-neo)] = ‘C920’ OR SUBSTRING([Und COD (non-neo)],1,2) = ‘I0’    OR SUBSTRING([Und COD (non-neo)],1,2) = ‘I2’ OR SUBSTRING([Und COD (non-neo)],1,2) = ‘I3’ OR SUBSTRING([Und COD (non-neo)],1,2) = ‘I4’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘I50’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘I51’ OR SUBSTRING([Und COD (non-neo)],1,2) = ‘I6’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘I70’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘I71’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘I72’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘I73’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘I74’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘I75’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘I76’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘I77’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘I78’    OR SUBSTRING([Und COD (non-neo)],1,2) = ‘J1’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘J40’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘J41’    OR SUBSTRING([Und COD (non-neo)],1,3) = ‘J42’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘J43’ OR SUBSTRING([Und COD (non-neo)],1,3) = ‘J44’)GROUP BY  [LA Code now],  [LA Name],  Sex,  [FiveYearTo85] ,  SUBSTRING([Und COD (non-neo)],1,3)ORDER BY  [LA Code now],  Sex,  [FiveYearTo85],  SUBSTRING([Und COD (non-neo)],1,3)
Numerator: aggregation /allocation Data was extracted from the mortality database as a text file, imported into Excel® and aggregated using the pivot table function to give groups comparable to the relative risk table (see below)
Numerator data caveats There may be some undercounting of calendar year deaths using registered death data because of underlying cause of death not being classified e.g. waiting for inquest results. Data were pooled across the 3 years
Denominator data caveats Data were pooled across the 3 years
Methods used to calculate indicator value We calculated age-specific attributable deaths as follows:

  1. Using the SAMMEC table below we calculated smoking attributable fractions for each disease-age-sex group using England smoking prevalence estimates.
Male Female
Disease Category Current Smoker Former Smoker Current Smoker Former Smoker
Malignant Neoplasms
Lip, Oral Cavity, Pharynx 10.89 3.4 5.08 2.29
Esophagus 6.76 4.46 7.75 2.79
Stomach 1.96 1.47 1.36 1.32
Pancreas 2.31 1.15 2.25 1.55
Larynx 14.6 6.34 13.02 5.16
Trachea, Lung, Bronchus 23.26 8.7 12.69 4.53
Cervix Uteri 0 0 1.59 1.14
Kidney and Renal Pelvis 2.72 1.73 1.29 1.05
Urinary Bladder 3.27 2.09 2.22 1.89
Acute Myeloid Leukemia 1.86 1.33 1.13 1.38
Cardiovascular Diseases
Ischemic Heart Disease
Persons Aged 35–64 2.8 1.64 3.08 1.32
Persons Aged 65+ 1.51 1.21 1.6 1.2
Other Heart Disease 1.78 1.22 1.49 1.14
Cerebrovascular Disease
Persons Aged 35–64 3.27 1.04 4 1.3
Persons Aged 65+ 1.63 1.04 1.49 1.03
Atherosclerosis 2.44 1.33 1.83 1
Aortic Aneurysm 6.21 3.07 7.07 2.07
Other Arterial Disease 2.07 1.01 2.17 1.12
Respiratory Diseases
Pneumonia, Influenza 1.75 1.36 2.17 1.1
Bronchitis, Emphysema 17.1 15.64 12.04 11.77
Chronic Airway Obstruction 10.58 6.8 13.08 6.78
  1. The calculation was performed using the formula

SAF = [(p0 + p1(RR1) + p2(RR2)) – 1] / [p0 + p1(RR1) + p2(RR2)]Where p0 = prevalence of non smokers, p1 = prevalence of current smokers; p2= ex-smoker prevalence; RR1 = relative risk in current smokers and RR2 = relative risk in former smokers.

  1. We aggregated the counts of deaths by relevant ICD codes,  age and sex groups
  2. We assumed that within broader age bands relative risks were constant
  3. We multiplied counts in 5-year age bands by the appropriate SAF to give smoking attributable deaths in each age-sex band.

We calculated directly age standardised rates using the batch calculator available from the erpho website http://www.erpho.org.uk/topics/tools/rates.aspx#12474

Small Populations: How Isles of Scilly and City of London populations have been dealt with We excluded the Isles of Scilly and the City of London from local authority calculations but included them in London GOR figure and Cornwall county estimate respectively.
Disclosure Control Not relevant
Confidence Intervals calculation method Confidence intervals have been calculated to exact Poisson limits using erpho “Template for producing funnel plots for directly standardised rates” (See resource “Quick Link ID” 12476 from www.erpho.org.uk) Method source: Spiegelhalter D. Funnel plots for comparing institutional performance. MRC Biostatistics Unit [cited 2004 Sept. 26]; Available from: URL:http://www.mrcbsu.cam.ac.uk/BSUsite/AboutUs/People/davids/funpap.pdf

Posted in: Indicators