Thomas D Koepsell

Quit Smoking Magic

Quit Smoking Cigarette

Get Instant Access

If epidemiology is the branch of science that seeks to understand the determinants of disease occurrence in populations, and if one such determinant is the effectiveness of organized efforts to prevent disease and promote health (Terris 1992a), then evaluation of these programs is a form of applied epidemiology.

This chapter concerns controlled evaluations of interventions aimed at entire communities in order to prevent disease or promote health. For present purposes, a community trial is defined as a study that involves at least one intervention site in which a community-wide health promotion or disease prevention program is implemented, and at least one control site without such an intervention that is studied concurrently, whether or not randomization is used to assign communities to treatment groups.

The design and analysis of community trials involves dealing not only with many of the generic methodological and practical issues that arise in most program evaluations but also with several special challenges that follow from aiming an intervention at an entire community rather than at selected individuals. The goal of this chapter is to orient the reader to these special issues, with an emphasis on principles and practical implications. Entry points will be provided into a rapidly growing and widely dispersed literature on evaluation of community interventions, where additional depth and details about these topics can be found.

Rationale for Community-Level Intervention

The high level of interest and investment in community trials over the last two decades has been fueled largely by the theoretical appeal of interventions aimed at intact social groups for prevention of disease and promotion of health. Because intervention at the community level has major implications for evaluation design, it will be helpful to review some of the key ideas behind this strategy.

Targeting Everyone May Prevent More Cases of Disease Than Targeting Just High-Risk Individuals

Rose (1985, 1992) has distinguished between two broad approaches to disease prevention in populations. Under a "high risk" strategy, risk-factor information about each individual is used to identify persons with the greatest chance of developing a preventable condition, and prevention efforts are then focused on those high-risk individuals. Under a "population" strategy, a preventive intervention is aimed at everyone in an attempt to produce a favorable shift in the overall risk-factor distribution in the population. Rose notes that for many diseases, the cases that occur in "high-risk" individuals may be only a small proportion of the total. Most fatal cases of coronary heart disease, for example, occur in people with "normal" cholesterol levels (Rose 1992), in whom dietary modification aimed at reducing cholesterol may lower the risk of coronary heart disease. Thus, a program aimed at both high- and low-risk individuals has the potential to prevent more cases than one aimed solely at high-risk individuals.

Environmental Modifications May Be Easier to Accomplish than Large-Scale Voluntary Behavior Change

It has long been recognized that community environment is an important determinant of disease risk (Terris 1992b). Modifying the physical, social, or legal/regulatory environment in which people live can be sometimes be more expedient and reach mgre people than attempting to induce voluntary behavior change on a mass scale. For example, requiring that children's sleepwear be manufactured from fire-retardant fabric may be a better way to prevent burns than teaching parents and small children about how to avoid ignition of sleepwear made from flammable fabric. Public or institutional policies, taking such forms as taxes on tobacco products, minimum legal drinking age, or subsidy of immunization services, can also create strong incentives or disincentives related to risk behavior (McKinlay 1993).

Risk-Related Behaviors Are Socially Influenced

Social learning theory holds that behavioral change is more readily achieved and maintained if norms and behavior in the peer group support the change

(Farquhar 1978). Community substance-abuse prevention programs, for example, have sought to change norms about illegal drug use among teenagers in order to influence individual behavior (Pentz et al. 1989).

Some Intervention Modalities Are Unselective by Nature

Fluoridation of the water supply to prevent dental caries automatically affects nearly everyone in a community. The word "mass" in "mass media" denotes the nonselective nature of public information dissemination that may be intended to change health-related attitudes and behavior (Flay 1987). Media campaigns about health topics typically seek to reach as large a share of the population within a media market as possible and can thus be regarded as community-level interventions.

Community Interventions Reach People in Their "Native Habitat"

Farquhar (1978) noted that people do not live in their doctors' offices, nor in the kinds of specialized environments created by large-scale, clinic-based lifestyle intervention trials such as the Multiple Risk Factor Intervention Trial (1982). Community interventions, in contrast, generally use intervention methods that apply in the "real-world" context of homes, workplaces, and neighborhoods.

Community Interventions Can Be Logistically Simpler and Less Costly on a Per-Person Basis

In contrast to the "high-risk" strategy described by Rose (1985), an intervention aimed at everyone in the community obviates the need to sort the population first into risk groups. Elimination of this step alone can reduce program complexity and cost (Farquhar 1978; Murray and Short 1995).

Examples of Community Trials

Community trials have involved a remarkable variety of settings, target populations, intervention strategies, outcomes, and approaches to evaluation. Four examples should help to illustrate this wide range of variation and provide specific contexts for later discussion of evaluation design issues.

The Community Intervention Trial for Smoking Cessation (COMMIT) sought to reduce the prevalence of heavy cigarette smoking in selected communities throughout the United States and Canada (Community Intervention Trial for Smoking Cessation [COMMIT] 1995a,b). In 1986, the National Cancer Institute invited applications from pairs of communities, requiring pair members to be from the same state or province and matched on approximate size and sociodemographic factors. Eleven pairs were chosen and randomized within pairs to intervention or control groups. Each intervention community received an average of $220,000/year for 4 years to mount a multifaceted program including public education through the media and community-wide special events, involvement of health care providers, activities at work sites and other community organizations, and resources to aid smokers in quitting. A process evaluation measured the nature and intensity of intervention activities of each type and smokers' exposure to them. An outcome evaluation used telephone surveys to monitor quitting in a cohort of about 1,100 smokers in each intervention and control community. Cross-sectional samples of smokers were also surveyed at baseline and follow-up to measure changes in smoking prevalence. The results showed no significant difference in quit rates among heavy smokers, but the quit rate among light to moderate smokers was about three percentage points higher in intervention sites compared to their matched controls (p = 0.004).

A randomized, double-blind, placebo-controlled community trial of vitamin A supplementation was carried out in rural Nepal in 1989-1990 (West et al. 1991). Twenty-nine local development units were studied, each with nine administrative wards, including about 28,000 children under 5 years of age. After obtaining written informed consent from the chairman of each development unit, the 261 wards were randomly allocated to treatment groups. Every 4 months, trained fieldworkers in each ward visited homes containing children under 5 years of age. In the intervention wards, children received a capsule containing 60,000 |xg of retinol equivalent; in control wards, the children received an identical-appearing capsule containing 300 |xg of retinol equivalent. Child deaths were identified both through vital records and at the time of 4-monthly follow-up visits. The trial, originally planned for 2 years, was halted early when results showed a statistically significant 30% reduction in child mortality in the vitamin A supplementation group.

In 1986, a community cardiovascular disease prevention program called Heart to Heart was mounted in the town of Florence, South Carolina (Goodman et al. 1995). The multicomponent intervention included media campaigns about smoking, physical fitness, and diet; public nutrition classes; information in supermarkets about food labeling; distribution of self-help resources for smoking cessation and weight control; a restaurant menu-labeling program; cholesterol and blood pressure screening at health fairs; and other activities. The town of Anderson, South Carolina, 200 miles away, served as a control community in which no special intervention activities were mounted. Telephone and questionnaire surveys were conducted in both sites in 1987 and 1991 to assess awareness of program components, knowledge about heart disease prevention, and key behavioral risk factors. A sample of individuals from each community also came to study clinics for measurement of blood pressure, lipid levels, and anthropometry. Although the two communities had been assigned en bloc to be intervention or control sites, individual respondents were used as the units of analysis. The results suggested modest but statistically significant benefits on the prevalence of high cholesterol and obesity, while changes in the prevalence of hypertension significantly favored the control community.

A publicly sponsored nurse-midwife program for low-income pregnant women was instituted in Boulder County, Colorado, to increase access to prenatal care and prevent adverse pregnancy outcomes (Lenaway et al. in press). To evaluate the program, birth-certificate data were analyzed for all singleton babies born alive to indigent mothers in Boulder County and in two neighboring control counties during a 16-month period. A random-effects logistic regression model was used to account for community-level allocation. The results showed significant reductions in the proportion of women who received inadequate prenatal care and in the proportion of infants with low 5-minute Apgar scores, and borderline reductions in the frequency of prematurity and low birth weight.

Key Terms and Concepts

Although most of this chapter will focus on more technical aspects of evaluation design, it is useful to begin by considering three broad issues that often arise in the context of community intervention studies and that can have major impact on program evaluation. The first two concern aspects of what we mean by a "community intervention."

What Is a Community?

Nutbeam (1986) defined a community as "A specific group of people usually living in a defined geographical area who share a common culture, are arranged in a social structure and exhibit some awareness of their identity as a group." Two features of this definition are worth noting. First, it places no restrictions on group size: groups as small as families or as large as nations could thus be regarded as "communities" of sorts. Many of the special methodological issues in evaluation of community interventions stem from allocation of intact social groups to different treatment conditions and do not depend strongly on group size. In practice, however, community interven tions have typically been aimed at social groups ranging in size from workplaces to counties, as illustrated by all four of the examples above. Second, sharing a common culture, social structure, and awareness of group identity implies some degree of similarity and connectedness: Members of a community share certain characteristics and influence each other. This simple observation has important ramifications for design and analysis of community interventions, because it implies that measurements taken on different members of the same community are not necessarily statistically independent.

Who Controls the Intervention?

The "intervention" part of the term "community intervention" refers to a defined plan of action. A critical issue for evaluators, however, is, Who controls the action plan? Table 6-1 describes two poles of a continuum. At one extreme, a community intervention may be part of a grand social experiment, in which the evaluator/experimenter (and/or the funding agency) is in control. Major elements of the program—the focal risk behaviors or health conditions, specific target communities, intervention modalities, nature and extent of resources, timing, and duration—are determined by the goals of the research and by people who are not indigenous community leaders. The Nepalese vitamin A trial was near this end of the spectrum. The primary goal of such a social experiment is to gain generalizable knowledge about the effects of a certain kind of intervention in certain kinds of communities in order to permit future application of that knowledge in other similar communities. Direct benefit to study communities themselves may be a bonus, but it is of secondary importance. Some communities participate in such an experiment as

Table 6-1. Two Models of Community Interventions

Social Experiment

"Grass Roots" Program


Outside community

Inside community



Internal or external

source sought out by


Primary goal

Generalizable knowledge

Solution of a perceived

about effectiveness

problem in the target


Control over nature,


Community leaders

timing, and target

or funding agency


Number of invention

One or more

Usually just one


controls, gaining no direct benefit from participation besides sharing in the knowledge that results when the study is completed.

At the other extreme, an intervention may be community based in a very different sense. It may have "grassroots" origins, arising from the felt needs and priorities of people in a particular community. The primary goal of the intervention is to prevent or solve some perceived problem in that community. Resources required to mount the intervention are raised within the community itself through public funds or voluntary contributions, or help is sought from an external service-oriented government agency or philanthropy known to support that kind of program. Control over the program remains within the community; the evaluator is an observer, not an experimenter. The Boulder County nurse-midwife program fell near this end of the spectrum.

Many gradations between these extremes are possible. In the COMMIT study, for example, the National Cancer Institute set conditions under which communities could participate, controlled allocation to treatment groups, and established general guidelines for acceptable intervention programs. Nonetheless, tailoring and implementation of the action plan in each site was directed by a community board that could choose from a menu of intervention options and that had considerable flexibility in adapting them to local conditions and needs. Participating communities also supplemented COMMIT funding with contributed time and other resources generated from within.

Degree of external control over the intervention plays a major role in determining what kind of study design is feasible for program evaluation. Large-scale social experiments like COMMIT lend themselves to having multiple communities in the intervention and control groups and random allocation of communities to treatment groups. Programs with "grassroots" origins in a single community must typically be evaluated using less elaborate study designs and without the benefits of randomization.

Was this article helpful?

0 0
Cigarette Crusher

Cigarette Crusher

Get All The Support And Guidance You Need To Permanently STOP Being A Slave To Nicotine And Cigarettes. This Book Is One Of The Most Valuable Resources In The World When It Comes To Easy Ways To Eliminate Smoking Addiction And Revitalize Your Body.

Get My Free Ebook

Post a comment