Statistical Analysis

This chapter includes background information and descriptions of the following tools FHOP
has developed to assist local health jurisdictions in their planning activities:

• Guidelines for Statistical Analysis
• Data Templates
• Basic Trend Analysis
• Guidelines for the collection and reporting of race and ethnicity data
• Epi BC
• Epi HOSP
• FHOP website

GUIDELINES FOR STATISTICAL ANALYSIS
OF PUBLIC HEALTH DATA WITH
ATTENTION TO SMALL NUMBERS

Many consumers of public health data have a limited understanding of statistics and scientific
methods. When these consumers are policy makers, they often are pressed to make decisions,
usually without as much information as they would like . In such situations, the temptation to
rely on weak data, rather than no data, is strong. One of the most common ways of pushing
data beyond its limits is to draw conclusions from rates based on a small number of events.
Relying on small numbers for analysis may be appropriate as long as the accompanying risks
are recognized and alternative approaches are appreciated.

The FHOP Technical Work Group, an advisory body with representatives from health-related
state and local agencies developed Guidelines for Statistical Analysis of Public Health Data with
Attention to Small Numbers. The guidelines are designed to assist local health jurisdictions and
health programs in applying clear and consistent approaches to the analysis and presentation of
data on health status and outcomes, with particular attention to situations involving small
numbers of cases or events. The simplified analytic techniques are intended for use by program
and data managers who may not have had training in statistics and who may have limited
access to epidemiologists and biostatisticians.

The basic document presents core information on standard measures used in public health and
how they are compared; describes the meaning and use of confidence intervals; and offers
standard formulas for the following: 1) Calculating confidence intervals for simple counts of
cases or events, for proportions, and for rates; 2) Calculating confidence intervals for differences
between proportions and between rates; and 3) In some situations, calculating confidence
intervals for ratios of proportions and ratios of rates. Special attention is given to the problems
of small numbers by description of methods for:

• Analyzing rates when dealing with small numbers
• Aggregating data across multiple years
• Aggregating data by geographic clustering
• Generating listing reports and conducting case studies

The table of contents from the guidelines and a summary page on recommended methods is
included as Appendix VII-A.

DATA TEMPLATES: TOOLS FOR
MONITORING CORE MCH INDICATORS
AT THE LOCAL LEVEL

FHOP health monitoring tools include a set of automated “data templates” for recording data
on the core MCH indicators. These templates are intended to standardize and simplify the
reporting of maternal and child health data used for local and state MCH needs assessments
and planning.

Using Excel spreadsheet software, the templates provide a standardized way of assembling,
reporting and analyzing data for selected indicators (such as low birth weight). Each template
has a specified reporting format (number, percent or rate), outcome measure (definition),
population of interest, definition of numerator and denominator , and age category (where
appropriate). When applicable, the templates provide benchmark targets from the Healthy
People 2010 Objectives. The templates contain up to eleven years of statewide data (in some but
not all indicators) for California. The user enters local numbers for the same years, and the
templates generate rates or percents for comparison, along with confidence intervals. The
current version of the templates allows for alternative calculations when numbers for the local
area are too small for standard statistical methods. The templates automatically generate
graphs comparing state and local trends over time. Data sources are specified and additional
recommended tables are listed for each template. Accompanying instructions guide the data
entry process, which is user-friendly.

Copies of the templates and much of the relevant data are available from FHOP’s website
for the following MCH indicators:

• Crude Birth Rate
• Distribution of Births by Race/Ethnicity, Trends in Distribution
• Fertility Rates (women ages 15-44)
• Teen Births (ages <15, 15-17, 18-19)
• First Trimester Initiation of Prenatal Care, Prenatal Care Adequacy
• Low Birth Weight, Very Low Birth Weight
• Breastfeeding Intent at Hospital Discharge
• Infant, Neonatal, Post-Neonatal, Fetal Mortality
• Injury Hospitalizations of Children: Unintentional, Assault, Self-Inflicted
• Fatal Injuries of Children: Unintentional, Suicide, Homicide
• Fatal Injuries of Youth (ages 15-19): Suicide, Homicide
• Fatal Drownings of Children (age 1-4)
• Fatal Injuries of Children and Youth (ages 0-24): Unintentional, Suicide, Homicide
• Domestic Violence Calls for Assistance, Weapons Related Calls, Arrests, Hospitalizations, Deaths

Appendix II-H provides an example of one of the data templates with sample local data.

DO WE HAVE A TREND?
A BEGINNERS GUIDE TO ANALYSIS OF TRENDS
IN COMMUNITY HEALTH INDICATORS

As detailed in previous sections, FHOP has developed sets of guidelines for the statistical
analysis of data, as well as templates for examining changes in key MCH indicators. These
resources, however, do not discuss approaches to trend analysis. Few public health managers
have the analytic expertise to determine whether a trend may be occurring and, if so, whether it
is statistically significant. This document was developed to assist local health jurisdictions in
making such determinations. We advise that it be used in concurrence with the previously
described Excel Data Templates and the Guidelines for Statistical Analysis of Public Health Data
with Attention to Small Numbers. Do We Have a Trend? describes some alternative ways of
identifying and analyzing trends, and provides references for trend analysis methods.

Do We Have a Trend? begins by discussing approaches to trend analysis. Subsequent sections
include definitions of technical terms, examples of simple and complex trends and instructions
on how to distinguish between and interpret them. The document’s Appendices include
instructions for using the Data Templates to calculate trends as well as other statistical
information.

Do We Have a Trend? is available for download from the FHOP website

GUIDELINES ON RACE/ETHNICITY DATA
COLLECTION, CODING, AND REPORTING

California has experienced tremendous demographic change in recent decades due primarily to
the surge of immigrants from around the world. Since the early 1990’s nearly half of the births
in California have been to foreign-born women. An increasing number of cultures with
differing health needs and beliefs demands more detailed data collection to help meet the health
challenges of each community in a culturally sensitive and competent way. In addition, the
federal government significantly revised methods for collecting and reporting data on race, as
reflected in the 2000 Census. The most significant of these revisions was the inclusion of multiracial
identifications. This change alters not only data collection, but presents analytic issues as
well.

To address the needs posed by California’s diverse population and the changes in government
data collection and reporting methodology, FHOP and its Technical Work Group have
developed the Guidelines on Race/Ethnicity Data Collection, Coding, and Reporting. The Guidelines
are intended to assist programs in bringing their race data collection in line with the 2000
Census and with the requirements of the Federal Office of Management and Budget (OMB).
The Guidelines also help to establish standards for the classification and reporting of race and
ethnicity in all publications of data from the California Department of Health Services.

The Guidelines consist of several major sections:

• Introduction
• Proposed Race/Ethnicity Guidelines – Basic Policy Summary
• Data Collection – Detailed Recommendations
• Data Coding
• Data Reporting/Tabulation
• Attachments and Appendices

These sections stress appropriate data collection methodologies and definitions of “minimum”
categories of race and Hispanic origin that comply with State law and Federal standards.
Beyond these standards, the Guidelines suggest additional data detail that may be used to
address their unique needs.

The latest version of the Guidelines will be presented to the California Department of Health
Services (CDHS) Executive staff for implementation in all CDHS programs. The revised
Guidelines are available for download at the FHOP website.

EPI BC 2002:
PUBLIC DOMAIN SOFTWARE
FOR ANALYSIS OF BIRTH CERTIFICATE DATA

Local maternal and child health directors are not always able to use their birth data in a timely
manner, either because they cannot access birth record data sets or because they lack easy-touse
analytic tools. FHOP developed Epi BC to improve access to and utilization of birth
certificate data in a standard format for comparisons within and across local health jurisdictions
and comparisons to the state. Epi BC 2002 is a user-friendly software program, based on the
Centers for Disease Control and Prevention (CDC) public domain software, Epi 2002, for
importing, reviewing, reporting, graphing, and mapping data collected from birth certificates.
The program allows local MCH Directors to analyze their own birth data for needs assessment
purposes.

The first version of Epi BC (originally named Epi MCH/BC) was created in mid-1994 and was
well received by local health jurisdictions in California. CDC subsequently released Epi2000 and
then Epi2002, Windows version of Epi INFO. This allowed FHOP to update Epi BC to a
Windows environment. The latest version, Epi BC 2002, is free to local health jurisdictions.
Advantages include:
• An Import format that allows any state or local health jurisdiction to import its data
• Variable names that are consistent with those used by the National Center for Health
Statistics (NCHS)
• Much improved graphics comparable to Excel
• A scaled down version of Arc View for easy mapping of data

The updated version of Epi BC is a powerful tool for analysis of birth certificate data. We hope
that it will become an integral part of local needs assessment and program planning activities.

The software and manual are available online from FHOP at
FHOP provides telephone and email support and training for users associated with MCH
programs in California local health departments.

EPI HOSP: PUBLIC DOMAIN SOFTWARE
FOR ANALYSIS OF HOSPITAL DISCHARGE DATA

There are few population-based data sources for monitoring child health morbidity. Hospital
discharge data are now collected by the majority of states and submitted to a federal agency in a
consistent format. The data sets are rich sources of information on childhood morbidity and are
useful for:

• Monitoring health status and outcomes
• Monitoring and evaluating the impacts of changes in health care delivery
• Injury surveillance
• Identifying patterns and trends in pregnancy related and ambulatory care sensitive
conditions

Epi HOSP is a software package for reviewing, analyzing, reporting, graphing, and mapping
key variables from the hospital discharge data for children less than 20 years of age. The
software tool enables local MCH staff to analyze their jurisdictions’ hospitalization data. Epi
HOSP is based upon the DOS version of Epi INFO, public domain software designed by the
Centers for Disease Control and Prevention. It uses Epi MAP to allow mapping of hospital
discharge variables.

To facilitate the importing of hospital discharge files into Epi HOSP, FHOP preprocesses hospital
discharge data for local health jurisdictions in California. A programmer’s manual is
available to assist others in preprocessing data from other states. With Epi HOSP, age groups
are defined; ambulatory care sensitive diagnoses and intentional or unintentional injuries are
flagged; new variables are created for grouped ICD-9-CM codes; charges are rounded and
sources of admissions are grouped together. Preprocessed hospital discharge data files for
children ages 28 days through 19 years are available for years 1992-99.

FHOP provides telephone and email support and training for users associated with MCH programs in
California local health departments.

In 2003, FHOP will release the windows version of Epi HOSP 2002, using CDC’s Epi 2002.