PHE 423 Module 6 Analysis Short Paper Example

Ninety Percent Still Sounding: Analyzing the First Year of the Smoke Alarm Program

[Student Name]

Southern New Hampshire University

PHE 423: Evaluation Methods in Public Health

Module Six Short Paper

[Instructor Name]

[Date]

The organization, setting and figures below are a composite written as a model document. No real employer, client, colleague or patient is described.

Ninety Percent Still Sounding: Analyzing the First Year of the Smoke Alarm Program

The evaluation plan for the river city's smoke alarm program set four questions: how many eligible homes the program reached, whether crews installed alarms as intended, how many served homes had a working alarm a year later compared with the neighborhood still waiting, and how many households had an escape plan. This paper reports the first year's data for each, with the uncertainty around the main results and the limits that readers should keep in mind.

Reach

Crews knocked on all 1,140 eligible rental homes in the two program neighborhoods and installed alarms in 712, a reach of 62%. Of the 428 homes not served, 268 never answered after three visits, 104 households declined and 56 landlords refused permission. The pattern matters. Homes that never answered were concentrated on two streets of duplexes where tenants turn over quickly, which are also the homes fire officials consider at highest risk. Glasgow et al. (1999) warned that a program's impact is the product of its reach and its effect, and a program that misses its riskiest homes can look better in outcome data than it is in the community.

Fidelity

Inspectors spot-checked 71 installations, about 10% of the total. In 64 of them, or 90%, an alarm was mounted on every sleeping level as the protocol required. The seven shortfalls were all in two-story homes where the crew installed a single alarm on the ground floor. Crew logs showed the safety talk was recorded as given in 689 of 712 homes. The evaluation cannot confirm what was said in each talk, a limit of relying on self-reported logs.

Working Alarms at Twelve Months

Health department staff revisited random samples of homes twelve months after installation. They reached 168 of 200 sampled served homes and 131 of 150 sampled comparison homes. Table 1 gives the results.

Table 1. Working smoke alarms at baseline and twelve months

Note. Intervals use the normal approximation for proportions.

The two groups began within two points of each other, and at twelve months the program neighborhoods were 44 points ahead, with an interval that stays far above the ten-point difference the plan set as meaningful. The comparison neighborhood changed little, which suggests that the gain is not explained by something affecting the whole city, such as news coverage after the fatal fires.

Why Some Alarms Stopped Working

Of the 17 served homes without a working alarm, 9 had removed it, most often from a kitchen ceiling where cooking set it off; 5 no longer had the alarm at all after a change of tenant; and 3 had damaged units. DiGuiseppi et al. (2002) found that free alarms in London did not reduce injuries, largely because many were never installed or were later taken down. Installing the alarms closes the first gap. The kitchen removals suggest that placement, not only installation, needs attention.

Escape Plans

In served homes, 61 of 168 households (36.3%) described a plan for getting out, compared with 29 of 131 (22.1%) in the comparison neighborhood. The difference of 14 points is smaller than for alarms and rests on a self-reported answer, so it should be read with more caution.

Threats to Validity

Three threats deserve attention. First, the 32 served homes that were sampled but never reached may differ from those that were. If none of them had a working alarm, the served figure would fall to 151 of 200, or 75.5%, still far above the comparison neighborhood, so the main conclusion holds even in that worst case. Second, the neighborhoods were not assigned at random, and a difference in tenants or landlords could contribute to the gap, although their similar baselines make a large bias unlikely. Third, staff knew which neighborhood they were visiting, but pressing a test button leaves little room for judgment.

Fire Injuries

The program neighborhoods recorded two fire injuries and no deaths during the year, against three injuries the year before. These numbers are far too small to show whether the program reduced harm, and the plan said in advance that they would not be used that way. Yellman et al. (2018) were able to estimate deaths and injuries prevented only by pooling several years of data from a large city program, which is the route the river city would need to follow.

Cost per Home with a Working Alarm

The program spent about $187,500 in its first year on alarms, crew time, outreach and supplies, or roughly $263 for each of the 712 homes served. Applying the twelve-month rate of 89.9% to all served homes gives about 640 homes with a working alarm, at roughly $293 each. That figure is not a measure of value, because it says nothing about fires prevented, but it gives the city a baseline for judging whether changes next year, such as evening visits to reach more tenants, raise or lower the cost of each working alarm. A later economic evaluation could build on the approach Yellman et al. (2018) used, once several years of injury data exist.

What Should Change Next Year

Three changes follow from the data. Crews should schedule evening and weekend visits on the high-turnover duplex streets, where most unanswered doors were. They should mount alarms at least ten feet from cooking appliances, or use photoelectric units near kitchens, to cut nuisance removals. And they should carry a second alarm on every visit so two-story homes get one on each sleeping level. Each change can be checked in next year's process data, since the third neighborhood will be measured the same way.

Conclusion

The program reached six in ten eligible homes, mostly installed alarms as intended and left nine in ten revisited homes with a working alarm after a year, against fewer than half in the neighborhood still waiting. Its biggest gaps are the homes it never reached and alarms placed where cooking sets them off. Both can be addressed before crews move to the third neighborhood.

References

DiGuiseppi, C., Roberts, I., Wade, A., Sculpher, M., Edwards, P., Godward, C., Pan, H., & Slater, S. (2002). Incidence of fires and related injuries after giving out free smoke alarms: Cluster randomised controlled trial. BMJ, 325(7371), 995-998. https://doi.org/10.1136/bmj.325.7371.995

Glasgow, R. E., Vogt, T. M., & Boles, S. M. (1999). Evaluating the public health impact of health promotion interventions: The RE-AIM framework. American Journal of Public Health, 89(9), 1322-1327. https://doi.org/10.2105/AJPH.89.9.1322

Yellman, M. A., Peterson, C., McCoy, M. A., Stephens-Stidham, S., Caton, E., Barnard, J. J., Padgett, T. O., Florence, C., & Istre, G. R. (2018). Preventing deaths and injuries from house fires: A cost-benefit analysis of a community-based smoke alarm installation programme. Injury Prevention, 24(1), 12-18. https://doi.org/10.1136/injuryprev-2016-042247