Spatial Clustering of Scleroderma in Three Michigan Counties Spatial Clustering of Scleroderma in Three Michigan Counties “The Toledo Twins” Sharon HensleyAlford Sarah Ann Cleveland
Background The disease that “turns people to stone” Classifications Chronic, connective tissue disease Unknown cause Collagen accumulation in some organs Classifications Localized Systemic
Who has scleroderma Approximately 150,000 people in the United States 4 times more women than men
Symptoms May Include Raynaud’s Phenomenon Swelling of hands and feet Pain and stiffness of joints Thickening of the skin Kidney, heart, and lung involvement Oral, facial, and dental problems
Diagnosis Difficult Involvement of several specialists May take months to years
Rates in Study Area Prevalence 242 cases per 1million adults Incidence 19 new cases per 1 million adults per year Reference: Prevalence, incidence and survival rates of systemic sclerosis in the Detroit metropolitan area. Mayes et al.
Age Distribution of Cases
Sex Distribution 134 Females 37 Males 3.62 times more females than males
Etiologies UNKNOWN Possibilities include: silica dust vinyl chloride monomer pet ownership some solvents appetite suppressants
Significance To Explore... Environmental associations Spatial pattern Disease process
Materials Disease Data Population Data Incidence data for 1989-1991 (N=171) Three counties: Macomb Oakland Wayne Population Data 1990 Census data Block and Tract divisions
Methods Gathering population data Geocoding cases Finding census data Extract pertinent information Geocoding cases ArcView Batch Match Digitizing unmatched cases
Working with Unmatched Cases Verifying address: US Post: www.usps.gov Using web map programs Mapquest: www.mapquest.com gives county lines Vicinity: www.mablast.com Digitizing
Matching Rate 161 Matched/166 Total 96% Matching Rate 171 Original Batch-Matched Addresses n=148 Non-matched n=18 Digitized n=13 Unable to Digitize n=5 5 Unmatchable Addresses n=166 161 Matched/166 Total 96% Matching Rate
Statistical Analysis Null Hypothesis: No spatial clustering Alter. Hypothesis: Spatial clustering Test statistic: Ipop, Moran’s I Statistical Program: STAT!
Macomb/Tracts STAT Output 4/13/1999 Assumption R Results from Ipop test Number of runs : 99 Ipop calculations Areas (m) : 194 cases (n) : 30 Population (x) : 724110 Ipop : -0.0000099 Ipop' : -0.2377846 E[I] : -0.0000014 % within : 99.9752702 % among : 0.0247298 Assumption R Variance : 0.0000000 z-score : -0.3175953 Significance : 0.7507920 (2-tailed) Approximation z-score : -0.3121641 Significance : 0.7549158 (2-tailed) Simulation Significance : 0.8800000 (2-tailed)
Macomb/Block STAT Output 4/14/1999 Results from Ipop test Number of runs : 99 Ipop calculations Areas (m) : 661 cases (n) : 30 Population (x) : 724112 Ipop : -0.0000208 Ipop' : -0.5008514 E[I] : -0.0000014 % within : 99.9957345 % among : 0.0042655 Assumption R Variance : 0.0000000 z-score : -0.3930764 Significance : 0.6942630 (2-tailed) Approximation z-score : -0.3855452 Significance : 0.6998335 (2-tailed) Simulation Significance : 0.6400000 (2-tailed)
Oakland/Tracts STAT Output 4/13/1999 Results from Ipop test Number of runs : 99 Ipop calculations Population (x) : 1101540 Ipop : 0.0000033 Ipop' : 0.0817093 E[I] : -0.0000009 % within : 99.8772695 % among : 0.1227305 Assumption R Variance : 0.0000000 z-score : 0.2045747 Significance : 0.8379044 (2-tailed) Approximation z-score : 0.2022084 Significance : 0.8397538 (2-tailed) Simulation Significance : 0.5600000 (2-tailed)
Oakland/Blocks
Wayne/Tracts STAT Output 4/13/1999 Results from Ipop test Number of runs : 99 Ipop calculations Areas (m) : 632 cases (n) : 86 Population (x) : 2159815 Ipop : -0.0000062 Ipop' : -0.1565116 E[I] : -0.0000005 % within : 100.4700628 % among : -0.4700628 Assumption R Variance : 0.0000000 z-score : -0.3578355 Significance : 0.7204664 (2-tailed) Approximation z-score : -0.3542334 Significance : 0.7231640 (2-tailed) Simulation Significance : 0.9200000 (2-tailed)
Wayne/Blocks
Discussion Limitations of Analysis Future Analysis Position uncertainty Residential history Reliability of census data Future Analysis Stratification by age, sex, race 3 county combination analysis Space/Time Analysis
First Honors: Andy Long Thank You First Honors: Andy Long Mark Wilson Dr. Mayes Geoff Jacquez