PU580 Unit 4 Dose-Response Analysis Example

Course	PU580 Environmental Epidemiology
Unit	Unit 4
Paper type	Dose-response analysis
Length	About 840 words, 6 pages double-spaced
Format	APA 7 student paper
School	Purdue University Global
Program	Master of Public Health (MPH)
Updated	October 2026

Free sample paper for PU580 Unit 4

The Curve Rises Once and Then Flattens: A Unit 4 Dose-Response Analysis Working a Composite Rubber Plant's Benzene Table for Trend, Threshold and the Four Artifacts That Produce a Plateau Without Any Biology

[Student Name]

Purdue University Global

PU580: Environmental Epidemiology

Unit 4 Dose-Response Analysis

[Instructor Name]

[Date]

The cohort of 5,300 workers and its table are composite; the exposure-response patterns, the attenuation literature and the research cited are real.

What this part is doingThe analysis treats the plateau as the finding to be explained rather than as the shape of the true relationship.

The Supplied Table

Cumulative exposure in parts per million-years, cases of acute myeloid leukemia, person-years, rate per 100,000 person-years, and rate ratio against the lowest band.

1. Under 10: 4 cases, 61,200 person-years, 6.5 per 100,000, reference.

2. Ten to 39: 9 cases, 38,400 person-years, 23.4, rate ratio 3.6.

3. Forty to 99: 7 cases, 21,100 person-years, 33.2, rate ratio 5.1.

4. One hundred and above: 5 cases, 14,700 person-years, 34.0, rate ratio 5.2.

Step One: Test for Trend Rather Than Eyeball It

A test for trend across the four ordered bands, using the midpoint of each as a score, gives a significant positive trend. That result is compatible with both a rising curve and a curve that rises and then flattens, so the trend test establishes that exposure matters and says nothing about shape. Reporting only the trend test, which many papers do, hides the whole question this analysis is about.

Step Two: Look at the Confidence Intervals Before the Point Estimates

The top two bands rest on 7 and 5 cases. Their intervals are wide enough to overlap each other and to overlap the second band's. The apparent plateau between bands three and four is therefore consistent with a continuing rise that the data cannot resolve, and the first thing to say about the flattening is that it may not exist. Five cases in the highest band cannot distinguish a plateau from a continued rise, and the plateau is the only interesting feature of the table.

Step Three: Four Artifacts That Flatten a Real Curve

1. Exposure misclassification that worsens with level. Historical estimates for the highest-exposed jobs usually come from the oldest records, often reconstructed from job titles rather than measurements, and error in the top band pulls its rate ratio toward the middle.

2. The healthy worker survivor effect. Workers who stayed long enough to accumulate the highest exposures are those who remained fit, and comparing them with lower-exposed workers who left introduces a selection that acts against the gradient.

3. Competing mortality. In a cohort old enough to have the highest cumulative exposures, death from other causes removes people before leukemia can occur, which lowers the observed rate in exactly the band where exposure is highest.

4. Saturation of the metric, not of the biology. Cumulative exposure multiplies intensity by duration, so a worker with 100 parts per million-years from two intense years and one with the same figure from twenty mild years are treated identically, and if intensity drives the effect the metric flattens the curve by construction.

Attenuation of exposure-response curves at high exposure is common enough in occupational cohorts that it has its own methodological literature, and these are the explanations it names (Stayner et al., 2003).

What this part is doingEach artifact is stated as a mechanism that would flatten the curve, so the reader can see that a plateau is weak evidence of a biological ceiling.

Step Four: Compare With the Published Curves

The classic cohort analysis of benzene and leukemia found a steep relationship with cumulative exposure and no evidence of a threshold at the levels studied (Rinsky et al., 1987). The large cohort in China, with better exposure reconstruction, found a dose-related increase across a wide range, including at lower levels than the early analyses had examined (Hayes et al., 1997). Neither supports a ceiling at the exposures in this composite table, which makes an artifact more likely than a biological plateau.

What this part is doingThe comparison with the published curves is kept short, since its only job is to show that a ceiling at these exposures is not what the literature describes.

Step Five: Is There a Threshold at the Bottom?

The lowest band, under 10 parts per million-years, carries a rate close to population background. That is compatible with a threshold and equally compatible with a linear relationship whose effect at low exposure is too small for four cases to detect. The analysis declines to claim a threshold, because a threshold claim from a lowest band with four cases is not supportable and because the regulatory consequence of the claim is large.

What the Analysis Concludes

1. Exposure and disease are related in this cohort, with a positive trend across bands.

2. The shape above about 40 parts per million-years cannot be determined from these data.

3. The plateau is more plausibly an artifact of exposure error, survivor selection, competing mortality or the cumulative metric than a biological ceiling.

4. No threshold should be inferred at the bottom of the range.

What Would Resolve the Shape

Intensity-weighted metrics alongside cumulative exposure, so a difference between high-intensity and long-duration exposure can be seen. Exposure lagged by 5 and 10 years, since leukemia latency makes recent exposure irrelevant. Analysis restricted to workers hired after measurement-based monitoring began, which trades sample size for exposure quality. And pooling with other cohorts, because 25 cases across four bands will not settle a question about curve shape however it is modeled.

References

Hayes, R. B., Yin, S. N., Dosemeci, M., Li, G. L., Wacholder, S., Travis, L. B., Li, C. Y., Rothman, N., Hoover, R. N., & Linet, M. S. (1997). Benzene and the dose-related incidence of hematologic neoplasms in China. Journal of the National Cancer Institute, 89(14), 1065-1071. https://doi.org/10.1093/jnci/89.14.1065

Rinsky, R. A., Smith, A. B., Hornung, R., Filloon, T. G., Young, R. J., Okun, A. H., & Landrigan, P. J. (1987). Benzene and leukemia: An epidemiologic risk assessment. The New England Journal of Medicine, 316(17), 1044-1050. https://doi.org/10.1056/NEJM198704233161602

Stayner, L., Steenland, K., Dosemeci, M., & Hertz-Picciotto, I. (2003). Attenuation of exposure-response curves in occupational cohort studies at high exposure levels. Scandinavian Journal of Work, Environment and Health, 29(4), 317-324. https://doi.org/10.5271/sjweh.737

PU580 Unit 4 instructions, in plain terms

Unit 4 usually supplies an exposure-response table and asks you to analyze it. Test for trend, then say explicitly that a positive trend is compatible with several shapes. Look at case counts and intervals in the highest bands before interpreting any plateau, because a flattening built on five cases may not exist. List the artifacts that can flatten a real relationship and explain the mechanism of each, including how a cumulative metric equates intensity with duration. Compare your table against published curves for the same agent. Decline to claim a threshold at the bottom unless the data can support one. Conclude about what can and cannot be determined, and say what would resolve the shape.

How the PU 580 Unit 4 example is put together

The analysis on this page reproduces the table with rates and rate ratios, runs a trend test, then puts the confidence intervals ahead of the point estimates. A highlighted line observes that five cases cannot distinguish a plateau from a continued rise. Four artifacts follow, each with its mechanism, and a margin note explains that a plateau is weak evidence of a biological ceiling. Published cohort curves are compared briefly under a further note. The threshold question at the low end is left open on four cases, and the conclusions are given as four numbered statements followed by the metrics, lags and pooling that would settle the shape.

Reading the PU580 Unit 4 grading rubric

Marking turns on whether the plateau is explained rather than reported. Credit begins with a trend test correctly interpreted as uninformative about shape. Substantial marks go to examining case counts and intervals in the top bands, since that is what reveals the apparent flattening may be noise. Each artifact named with its mechanism earns points, especially the distinction between saturation of a metric and saturation of biology. Comparing with published curves for the same agent is rewarded. Refusing a threshold claim from a sparse lowest band scores well, because the regulatory consequence is large. Conclusions stated as what can and cannot be determined complete the grade.

PU580 Unit 4 help from the desk

The commonest error is reading the curve off the point estimates, which treats five cases as if they were five hundred. The second is announcing a biological ceiling, when exposure misclassification, survivor selection, competing mortality and the metric itself all flatten curves. Third, do not claim a threshold because the lowest band looks like background, since four cases cannot distinguish a threshold from a small linear effect. Watch the cumulative metric, which equates two intense years with twenty mild ones. And lag the exposure, because recent exposure cannot have caused a disease with this latency. A worked version of your own table comes back inside two days. One further check worth adding: report the rate ratios against the lowest band and against an external population, since a cohort's own low-exposure group may itself be exposed.

Get a PU580 Unit 4 example written to your instructions

PU580 Unit 4 tables differ by agent and by how exposure was banded, so the analysis is built on the data your section supplies. Upload that table or dataset with the questions and rubric. Within 24-48 hours the free first custom sample reads each band with its denominator, tests the trend, and weighs whatever departs from it. The paper above is an original model document written by our desk, not a submitted student paper and not an official Purdue University Global document.

More PU580 papers and MPH samples

PU580 Unit 4 questions, answered

Where can I find a free PU580 Unit 4 sample paper?

The whole dose-response analysis, table and artifacts included, is published here.

What does a positive trend test prove?

That exposure matters. It says nothing about the shape of the relationship, which is the question a plateau raises.

Why might a curve flatten without biology?

Exposure error that grows with level, survivor selection, competing mortality, and a cumulative metric that treats intensity and duration as interchangeable.

Can a threshold be claimed here?

No. The lowest band holds four cases, which cannot separate a true threshold from a small effect the data are too sparse to see.

What would resolve the shape?

Intensity-weighted metrics, lagged exposure, analysis restricted to monitored hires, and pooling with other cohorts.