Population Surveys Methodologic problems: standardization Maura Pugliatti, MD, PhD Associate Professor of Neurology Dept. of Clinical and Experimental Medicine, Unit of Clinical Neurology University of Sassari, Italy 1st International Course of Neuroepidemiology Chisinau, Moldova, 24-28 Sept. 2012 Disease determinants: factors of importance for disease occurrence Personal characteristics Age Year of Birth Sex Marital status Occupation Education Income Ethnic group/Religion Time Short term variations Long term variations (secular changes) Place International variations Regional variations Local variations Urban/Rural variations Gallitto et al. Epilepsia, 2005. Standardization Standardization allows for comparisons when the population structures differ and is key in assessing the potential influence of environmental or cultural factors on disease rates in a region. A standardized estimate is a crude estimate that has been adjusted for differences in age composition between the study population and a standard (referral) population. Standardized estimate based on the same standard population are comparable and the ratio between two standardized estimates is called Standardized Rate Ratio (SRR) Standardization for gender and age There are two ways of computing standardized estimates: - Direct Standardization: it calculates a weighted average of the region’s age-specific estimates where the weights represent the age-specific sizes of the standard population. - Indirect Standardization: it uses age-specific estimates (eg., mortality) from the standard population to derive expected estimates (eg., deaths) in the region’s population. Choosing which formula to use will depend on what data you have access to. Type of data we can have access to: Table 1, US Population in thousands and State Age-Specific Mortality Rates Age 0-4 5-24 25-34 35-44 45-54 55-64 65-74 75+ Population US 18456 72053 43675 35264 24163 21830 17897 12470 Mortality New Mortality York Arizona 0,00272 0,00269 0,00066 0,00076 0,00192 0,00146 0,00321 0,00224 0,0054 0,00484 0,01212 0,01102 0,02793 0,02408 0,08753 0,07359 Direct Standardization (eg., incidence, prevalence, mortality) Table 2, U,S, Age-Specific Mortality and State Populations in Thousands Age 0-4 5-24 25-34 35-44 45-54 55-64 65-74 75+ US Death Rate 0,00251 0,00065 0,00135 0,0022 0,00486 0,01236 0,0273 0,08513 New York Population 1275 5023 3018 2572 1909 1784 1347 981 Arizona Population 299 1028 614 479 315 307 277 170 Indirect Standardization (eg., mortality) Direct Standardization: To compute the direct standardized mortality rate (SMR): 1. For each age group, you need to multiply Mr by Ps. 2. Add them up. 3. Divide the sum by the total standard population. 4. Multiply by 1000, or give the rate in terms of "per thousand population" SDR1 = [Sum age groups (Mr x Ps)]/Pt x 1000 Mr is the age-specific mortality rate for the region. Ps is the number of people in the age group in the standard population. Pt is the total standard population. Indirect Standardization: To compute the indirect standardized mortality rate: 1. For each age group, you need to multiply Ms by Pr. 2. Add them up. 3. Divide the sum into the number of deaths in the region. 4. Multiply by the crude death rate. SDR2 = Dr/[Sum age groups (Ms Pr)] x CDRs Ms is the age-specific mortality rate for the standard population. Pr is the number of people in the age group in the region’s population. Dr is the number of deaths in the region. CDRs is the crude death rate for the standard population. DIRECT METHOD OF STANDARDIZATION Examples Example: Table 1, US Population in thousands and State Age-Specific Mortality Rates Age Population Mortality New Mortality York 0,00272 0,00066 0,00192 0,00321 0,0054 0,01212 0,02793 Arizona 0,00269 0,00076 0,00146 0,00224 0,00484 0,01102 0,02408 NY AZ 0-4 5-24 25-34 35-44 45-54 55-64 65-74 US 18456 72053 43675 35264 24163 21830 17897 50,20 47,55 83,86 113,20 130,48 264,58 499,86 49,65 54,76 63,77 78,99 116,95 240,57 430,96 75+ 12470 0,08753 0,07359 1091,50 917,67 Tot 245808 2281,23 1953,31 SDR1 for New York = ??? SDR1 for Arizona = ??? SDR1 = [Sum age groups (Mr x Ps)]/Pt x 1000 Mr is the age-specific mortality rate for the region. Ps is the number of people in the age group in the standard population. Pt is the total standard population SDR1 for New York = 9,3 SDR1 for Arizona = 7,9 Prevalence of ALS in northern Sardinia (prov. of Sassari and Nuoro) in 2004 <35 35-44 45-54 55-64 65-74 75-84 85+ Tot M 97725 37116 31494 26850 19544 10467 2788 225984 M 1 0 4 8 4 3 0 20 SS 2004 F 1 1 2 5 3 0 0 12 M 1,0 0,0 12,7 29,8 20,5 28,7 0,0 8,9 SS 2004 F 93407 37579 31413 28367 23266 15599 5069 234700 SS 2004 F 1,1 2,7 6,4 17,6 12,9 0,0 0,0 5,1 T 191132 74695 62907 55217 42810 26066 7857 460684 M 56678 20607 17787 14750 11653 6700 1846 130021 M 0 2 5 5 4 1 0 17 NU 2004 F 0 0 0 1 7 3 0 11 T 2 1 6 13 7 3 0 32 T 1,0 1,3 9,5 23,5 16,4 11,5 0,0 6,9 M 0,0 9,7 28,1 33,9 34,3 14,9 0,0 13,1 NU 2004 F 53519 19978 17189 15798 14247 9962 3279 133972 T 0 2 5 6 11 4 0 28 NU 2004 F 0,0 0,0 0,0 6,3 49,1 30,1 0,0 8,2 TOT 2004 F 146926 57557 48602 44165 37513 25561 8348 368672 T 110197 40585 34976 30548 25900 16662 5125 263993 M 154403 57723 49281 41600 31197 17167 4634 356005 M 1 2 9 13 8 4 0 37 TOT 2004 F 1 1 2 6 10 3 0 23 T 0,0 4,9 14,3 19,6 42,5 24,0 0,0 10,6 T 2 3 11 19 18 7 0 60 M 0,6 3,5 18,3 31,3 25,6 23,3 0,0 10,4 T 301329 115280 97883 85765 68710 42728 12982 724677 The island of Sardinia, Italy 2004 census pop. (denominator) Sardinian provinces of Sassari and Nuoro 2004 number of ALS cases TOT 2004 F 0,7 1,7 4,1 13,6 26,7 11,7 0,0 6,2 T 0,7 2,6 11,2 22,2 26,2 16,4 0,0 8,3 2004 ALS prevalence Pugliatti M, Parish LD, et al. J Neurol, in press Prevalence of ALS in northern Sardinia (prov. of Sassari and Nuoro) in 2004 <35 35-44 45-54 55-64 65-74 75-84 85+ Tot M 0,4 0,0 1,7 3,6 2,0 1,6 0,0 9,3 8,9 M 11714767 4612114 3761710 3383481 2737383 1526720 332433 28068608 SS 2004 F 0,4 0,4 0,8 2,1 1,4 0,0 0,0 5,2 5,1 F 11247020 4579893 3851041 3609738 3282765 2433082 816098 29819637 T 0,4 0,2 1,3 2,8 1,7 0,8 0,0 7,2 6,9 M 0,0 1,6 3,8 4,1 3,3 0,8 0,0 13,6 13,1 T 22961787 9192007 7612751 6993219 6020148 3959802 1148531 57888245 NU 2004 F 0,0 0,0 0,0 0,8 5,4 2,5 0,0 8,6 8,2 WeightM 0,4173619 0,1643157 0,1340184 0,1205432 0,0975247 0,0543924 0,0118436 1 T 0,0 0,8 1,9 2,4 4,4 1,6 0,0 11,1 10,6 WeightF 0,3771682 0,1535865 0,1291445 0,1210524 0,1100874 0,0815933 0,0273678 1 M 0,3 0,6 2,4 3,8 2,7 1,6 0,0 11,2 10, 4 WeightT 0,396657 0,158789 Standard population 0,131508 2004 census pop. (denominator) 0,120806 ITALY 0,103996 0,068404 0,01984 1 TOT 2004 F 0,3 0,3 0,5 1,6 2,9 1,0 0,0 6,6 T 0,3 0,4 1,5 2,7 2,7 1,1 0,0 8,7 6,2 8,3 Standardized ALS prevalence in northern Sardinia Crude prevalence Pugliatti M, Parish LD, et al. J Neurol, in press Prevalence and Characteristics of Epilepsy in the Aeolian Islands Gallitto et al. Epilepsia, 2005. Prevalence and Characteristics of Epilepsy in the Aeolian Islands M F 5-14 42357 39261 15-24 40421 38737 25-34 36087 36298 35-44 31843 31191 45-54 26647 27514 55-64 22230 23907 65-74 14707 17861 75+ 10692 15151 Tot 224984 229920 5–14 15–24 25–34 35–44 45–54 55–64 65–74 >74 Total M 5,01 0 2,9 4,27 2,99 1,41 6,76 0 2,95 F 5,09 2,56 3,03 4,58 4,63 0 4,27 1,99 3,3 WeightM WeightF WeightT 81618 0,188267 0,170759 0,179418 79158 0,179662 0,16848 0,17401 72385 0,160398 0,157872 0,159121 63034 0,141535 0,13566 0,138565 54161 0,11844 0,119668 0,11906 46137 0,098807 0,10398 0,101421 32568 0,065369 0,077684 0,071593 25843 0,047523 0,065897 0,05681 454904 1 1 1 T T 5,05 1,25 2,96 4,42 3,75 0,73 5,41 1,24 3,13 Age-specific prevalence of epilepsy in Aeolian Islands, 1999 Gallitto et al, Epilepsia 2005 5–14 15–24 25–34 35–44 45–54 55–64 65–74 >74 Total M 0,94 0,00 0,47 0,60 0,35 0,14 0,44 0,00 2,95 F 0,87 0,43 0,48 0,62 0,55 0,00 0,33 0,13 3,29 1999 census pop. (denominator) Northern Sardinian population T 0,91 0,22 0,47 0,61 0,45 0,07 0,39 0,07 3,19 Standardized prevalence of epilepsy to the 1999 Sardinian pop. Gallitto et al. Epilepsia, 2005. Prevalence and Characteristics of Epilepsy in the Aeolian Islands Crude prevalence Stand. Prev to 2001 Italian pop. Stand. Prev to 1999 Sardinian pop. M F T 2,95 2,87 2,95 3,30 3,15 3,29 3,13 3,01 3,19 per 1000 Gallitto et al. Epilepsia, 2005. * * Gallitto et al. Epilepsia, 2005. DIRECT METHOD OF STANDARDIZATION Other Examples Direct standardization Example: Pop. 0-29 yrs 30-59 yrs 60+ yrs Total 3145000 3057000 1294000 7496000 Sweden Deaths Mortality 3523 10928 59104 73555 1,1 3,6 45,7 9,8 Pop. 741000 275000 59000 1075000 Panama Deaths Mortality 3904 1421 2456 7871 5,3 5,2 41,6 7,3 The choice of a standard population affects the comparison. Here the Sweden population is used as standard, the Panama rates applied to this population 0-29 yrs 30-59 yrs 60+ yrs Total 3145000*5,3/1000 3057000*5,2/1000 1294000*41.6/1000 SMR: SRR 86231,5*1000/7496000 11.5/9.8 16569,6 15796,4 53865,5 86231,5 11,5 1,17 Direct standardization Example: Pop. 0-29 yrs 30-59 yrs 60+ yrs Total 3145000 3057000 1294000 7496000 Sweden Deaths Mortality 3523 10928 59104 73555 1,1 3,6 45,7 9,8 Pop. 741000 275000 59000 1075000 Panama Deaths Mortality 3904 1421 2456 7871 5,3 5,2 41,6 7,3 Here the Panama population is used as standard, the Swedish rates applied to this population 0-29 yrs 30-59 yrs 60+ yrs Total 7411000*1.1/1000 275000*3.6/1000 59000*45.7/1000 SMR: SRR 4508*1000/1075000 7.3/4.2 830,1 983,1 2694,9 4508,0 4,2 1,75 Direct Method of Standardization Epilepsy: crude data from two studies 0-9 Age groups 10-19 20-39 40+ Tot Study A N. Cases At risk population Prevalence (per 1000) 62 10500 5,9 280 22500 12,4 131 25000 5,2 126 42000 3,0 599 100000 6,0 Study B N. Cases At risk population Prevalence (per 1000) 165 24500 6,7 538 42000 12,8 69 12500 5,5 68 21000 3,2 840 100000 8,4 35000 64500 37500 63000 200000 Standard population (sum of A+B) Standardized prevalence Study A Study B 7,0 per 1000 7,4 per 1000 =(5.9*35000)+(12.4*64500)+(5.2*37500)+(3.0*63000)/200000 =(6.7*35000)+(12.8*64500)+(5.5*37500)+(3.2*63000)/200000 Rosati G, Granieri E. Manuale di Neuroepidemiologia Clinica, 1990 Thank you