
CHP Methods and Sources of Data
Population
Data were collected on the six towns of the Lower Naugatuck Valley (Ansonia, Beacon Falls, Derby, Oxford, Seymour and Shelton), Bridgeport, Hartford and New Haven from publicly available data sources (e.g. the Department of Public Health). Specific demographics of these towns are available in subsequent sections of this document (see Population Statistics).
Assessment of the Previous Reports
The 1998, 2000, and 2003 Valley Health Profiles were reviewed to assess sections of the document that needed updating.
Data acquisition
The collection of data to update the Community Health Profile was conducted mainly via publicly available datasets. Data sources used in the previous report were contacted and electronic data were accessed through the Internet or hard copies were sent to the center for manual data reentry.
Data storage: Phone interviews, data collection, manipulation and presentation took place at the YaleGriffin Prevention Research Center in Griffin Hospital, Derby, CT under the supervision of David Katz, MD, MPH, and Veronika Northrup, MPH.
Data Analysis
Incidence and mortality data are presented in frequency tables, rates (per 100,000 people), and graphs. For trend analysis, rates of individual towns in the Valley, as well as total Valley rates were compared to rates of Bridgeport, Hartford, New Haven, and Connecticut, by examining confidence intervals around the rates (see Definitions of Rates and Terms). An overlap in confidence intervals indicated no statistically significant difference between rates. The purpose of this statistical testing is to establish whether two rates are truly different, or that there is a statistical chance that the rates are not different. That statistical chance is based on the existence of a random error in the calculation of the true rate. (Such error can come from a reporting error or a mistake in entering data). For example, if a rate is 100 with 95 percent of the time falling within the bounds of 89 and 111 interval, is that rate statistically different from a rate of 115, which 95 percent of the time falls within the bounds of 105 and 125? In this case, there is a chance that the first rate (given that a random error in the calculation of the rate exists) can be equal to 105, which is the number that falls within the bounds of the second rate’s true value. Therefore, the two rates are not statistically different. Caution should be taken in translating a statistical finding, or a lack thereof, into a significant finding. If a rare event, such as a rare disease, takes place in a small population, the magnitude of an incidence rate can fluctuate from one time point to another time point. However, a seemingly large difference between two incidence rates of a rare event in a small population may not be statistically significant based on the examination of the confidence intervals around each rate. A decision to establish a significant trend of some event should take into consideration a statistical significance testing, the nature of the event and the size of the population.
Definition of Rates and Terms

