PHLT 8520 Week 11 Final Methods Paper Example

Course	PHLT 8520 Advanced Epidemiological Methods
Week	Week 11
Paper type	Final methods paper
Length	About 1,150 words, 7 pages
Format	APA 7 student paper
School	Walden University
Program	PhD in Public Health
Updated	October 2026

Free sample paper for PHLT 8520 Week 11

One Design, Fully Specified: Final Methods for a Record-Linked Cohort of First Pregnancies Among Health-System Employees, Night Rosters in Weeks 4 to 20 and Preterm Birth, With Three Distortions It Cannot Remove

Student Name

Walden University

PHLT 8520: Advanced Epidemiological Methods

Instructor Name

Month Day, Year

What this page is doingThe title promises completeness and candor at once, a full specification and an admission of what remains, which is the balance a final methods paper must strike.

Purpose and Question

This paper specifies the methods for a study of whether working nights early in pregnancy changes the risk of delivering preterm. It gathers the decisions made over the term into a single protocol, gives the reason for each and closes with the distortions that survive every decision. The question is causal and narrow: among first pregnancies of women employed in rostered jobs at a composite nine-hospital health system, what is the effect of a schedule that includes any night shift during gestational weeks 4 through 20, compared with a schedule of days or evenings only, on the risk of a live birth before 37 completed weeks?

What this page is doingRestating the question in causal terms with every boundary attached tells the reader exactly which estimate the methods are built to produce.

The Target Trial and Its Emulation

The design is a record-linked cohort set up to emulate a trial that could never be run (Hernán & Robins, 2016). Table 1 places each element of that trial beside the cohort's version of it.

Table 1

Target Trial Elements and the Cohort's Emulation

Element	Hypothetical trial	Emulation in the cohort \|
Who is eligible	First singleton pregnancies, confirmed by week 6, in rostered employees	First singleton pregnancies confirmed in system clinics, January 2016 to December 2023, employee rostered from week 4 through week 20 \|
Strategies compared	Nights allowed in weeks 4-20 versus days or evenings only	At least one clocked night shift in the window, versus none \|
Assignment	Random	Adjustment for measured confounders, with a sensitivity analysis for one unmeasured \|
Time zero	Confirmation	Clinic confirmation date, with exposure counted from week 4 \|
Outcome	Live birth before 37 weeks	Live birth at 23 to 36 completed weeks, dated by first-trimester ultrasound where available \|
Causal contrast	Intention to treat	Strategy as recorded in timekeeping, with a per-protocol analysis weighted for leaving the strategy \|

Note. Composite health system.

Population and Linked Sources

Three record systems are joined by employee number inside the system's research warehouse: timekeeping, which logs each clocked shift with its start and finish; human resources, which holds title, grade, hire date and pay band; and the obstetric record, which supplies confirmation, ultrasound dating, obstetric history and delivery. Restriction to first births, chosen in Week 8, leaves about 3,400 pregnancies. The restriction removes parity and an earlier preterm delivery from the analysis entirely, since an earlier delivery may have been shaped by nights worked in that earlier pregnancy (Howards et al., 2012). A secondary analysis of later births, adjusted for obstetric history, will be shown and labeled as estimating a different quantity.

Comparison Group

Women whose schedules held no nights in the window are drawn from the same job titles as the exposed women. Holding title constant removes differences in training, physical demands and workplace hazards. The general population was set aside because employment itself selects for health, and evening-only staff were folded into the comparison group, with a sensitivity analysis that separates them, because late evenings carry some of the same circadian disruption.

Exposure and Its Accuracy

Night work is read from timekeeping, entered before any outcome is known. Diaries kept by 212 employees showed timekeeping catching 96 of every 100 truly exposed women and correctly clearing 95 of every 100 unexposed ones, so misclassification is small and, being unrelated to the outcome, is expected to understate the effect slightly. The corrected estimate will be reported beside the uncorrected one. Two dose measures, the number of nights and the longest consecutive run, will be analyzed in addition to the yes-or-no strategy.

Confounding Control

Four pre-exposure variables form the adjustment set read from the Week 3 diagram: job grade, pay band as a partial stand-in for household income, maternal age and chronic disease present before pregnancy. Sleep loss lies on the causal path and is not adjusted. Complications arising after week 4 are handled as time-varying, not as baseline covariates. Because complications can move women off night rosters, an approach Buckley et al. (2015) describe for the healthy worker survivor effect, the per-protocol analysis uses inverse probability weights for remaining on the assigned strategy, with complications entering the weight models week by week.

Analysis

Risk ratios come from log-binomial regression, with robust standard errors. The intention-to-treat contrast compares women by their recorded strategy at week 4 regardless of later changes; the per-protocol contrast weights women who stay on their strategy. Spontaneous and medically indicated preterm births are modeled separately. With roughly 3,400 first births, a modest risk ratio of 1.2 would be missed about a third of the time, a limitation stated in advance.

Planned Bias Analyses

VanderWeele and Ding (2017) recommend reporting the E-value for both the estimate and its confidence limit, and both will appear. Results of each analysis will be shown together, so a reader can see how much the estimate shifts when each concern is addressed in turn.

Table 2

Bias Analyses Reported With the Main Estimate

Concern	Analysis	Source of parameters \|
Financial strain from a second job	E-value and bounding factor across a grid of strengths	Disclosure records and published gradients \|
Household confounding	Partners' nights as a negative control exposure	540 dual-employee pregnancies \|
Exposure misclassification	Correction using diary sensitivity and specificity	Diary substudy \|
Gestational dating	Restriction to pregnancies dated by first-trimester ultrasound	Obstetric record \|
Losses before confirmation	Counting registry-detected early losses as a competing event	Pregnancy test records \|

Note. All analyses are prespecified.

Three Distortions That Remain

Uneven Ultrasound Dating

Early ultrasound dating covered 81% of night workers' pregnancies against 88% of day workers', and later dating tends to classify more births as preterm. This error is differential by exposure and would push the estimate upward. Restricting to early-dated pregnancies limits it but loses about one pregnancy in seven and may select women with better access to care.

Survival to Confirmation

Pregnancies lost before clinic confirmation never enter, and if night work raises very early loss while those losses share causes with preterm birth, the surviving pregnancies of night workers are less susceptible. This collider structure would pull the estimate down. Registry-detected losses capture only part of it.

Departures Before Conception

Women who could not tolerate nights and left before conceiving are absent, and their intolerance may share roots with obstetric risk. The night workers who remain are a hardier group. This also pulls the estimate down, and no analysis in the cohort can recover women who left.

What this page is doingGiving each remaining distortion a direction lets readers judge whether the net bias could have produced the result.

Reading the Result

Two of the three remaining distortions point toward the null and one away from it. A clearly elevated estimate that survives restriction to early-dated pregnancies would therefore be difficult to dismiss. An estimate near 1 would agree with the larger, payroll-based evidence (Specht et al., 2019), though the two downward biases mean a small harm could be hidden. The paper commits to reporting either result in the same form.

References

Buckley, J. P., Keil, A. P., McGrath, L. J., & Edwards, J. K. (2015). Evolving methods for inference in the presence of healthy worker survivor bias. Epidemiology, 26(2), 204-212. https://doi.org/10.1097/EDE.0000000000000217

Hernán, M. A., & Robins, J. M. (2016). Using big data to emulate a target trial when a randomized trial is not available. American Journal of Epidemiology, 183(8), 758-764. https://doi.org/10.1093/aje/kwv254

Howards, P. P., Schisterman, E. F., Poole, C., Kaufman, J. S., & Weinberg, C. R. (2012). "Toward a clearer definition of confounding" revisited with directed acyclic graphs. American Journal of Epidemiology, 176(6), 506-511. https://doi.org/10.1093/aje/kws127

Specht, I. O., Hammer, P. E. C., Flachs, E. M., Begtrup, L. M., Larsen, A. D., Hougaard, K. S., Hansen, J., Hansen, Å. M., Kolstad, H. A., Rugulies, R., Garde, A. H., & Bonde, J. P. (2019). Night work during pregnancy and preterm birth: A large register-based cohort study. PLOS ONE, 14(4), Article e0215748. https://doi.org/10.1371/journal.pone.0215748

VanderWeele, T. J., & Ding, P. (2017). Sensitivity analysis in observational research: Introducing the E-value. Annals of Internal Medicine, 167(4), 268-274. https://doi.org/10.7326/M16-2607

Reading the PHLT 8520 Week 11 instructions

Final papers in advanced methods courses usually ask students to assemble a complete methods section for their study, integrating the design, measurement, confounding, selection and sensitivity work completed during the term. Instructors expect each decision to appear with its reason and a citation, not as a list. Prompts often require a section on limitations that states the direction of remaining biases. Many ask students to show how the design emulates an ideal trial. Expect eight to twelve pages in some sections and fewer in others; tables summarizing the design and planned analyses are common. Graders look for consistency with earlier assignments and for a clear account of what changed and why. A statement on how results of either sign will be reported is often welcomed.

How the PHLT 8520 Week 11 example is put together

The paper opens by stating the causal question with every boundary attached. Table 1 places six target trial elements beside their emulation. Sections follow on population and linked records, including the restriction to first births and its reason; the comparison group; exposure and its measured accuracy; confounding control, including weighting for complications that move women off nights; and the analysis plan with a stated power limit. Table 2 lists five prespecified bias analyses with their parameter sources. Three subsections describe the distortions that remain, each with a direction and a reason. A closing section explains how the net direction of those distortions shapes the reading of a high or a null result.

Reading the PHLT 8520 Week 11 rubric

The final paper is scored on completeness, coherence and candor. Completeness credit comes from specifying every element a reader would need to repeat the study. Coherence credit comes from tracing each decision back to earlier reasoning, such as restriction to first births and the comparison group defense. The bias analysis table earns planning credit because each concern is matched to a method and a data source. Candor is the section graders read most closely: the sample names three distortions it cannot remove and gives each a direction, then explains how they combine. Prespecifying how a null and a positive result will be read earns further credit. Formatting, tables and citations take the remaining points. Final papers that omit limitations or describe them without direction rarely score at the top.

Common PHLT 8520 Week 11 mistakes, and how to avoid them

Final methods papers often read like a pile of weekly assignments. Rewrite instead of pasting: each decision should appear once, with its reason. Use a target trial table so readers can check the design quickly. Put validation and sensitivity work in a single table of planned bias analyses. Be specific about what you cannot fix, and give each remaining distortion a direction so readers can judge the net effect. Say in advance how you will report a null result. Check that numbers match across sections, since graders notice when sample sizes drift. Keep later revisions of your question consistent with the methods. Our writers will draft a free first methods chapter around your own study on request.

Get a PHLT 8520 Week 11 example written to your instructions

Bring the Week 11 paper prompt and rubric, the research question carried through your term and any earlier assignments on it. The paper specifies one design against its ideal protocol, carries your earlier decisions forward with their reasons, and ends on the distortions it cannot remove. First paper free; 24-48 hours.

More PHLT 8520 papers and PhD in Public Health samples

PHLT 8520 Week 11 questions, answered

Where can I find a free PHLT 8520 Week 11 sample paper?

A full Week 11 final methods paper is posted above, specifying a record-linked cohort of first pregnancies against its target trial, with two tables and real sources.

What should a final epidemiology methods paper include?

The causal question, design and emulation of an ideal trial, population and data sources, exposure and outcome definitions with their accuracy, confounding control, analysis plan, bias analyses and limitations with directions.

How should remaining biases be described?

Name each one, explain its mechanism and state its likely direction, so readers can judge whether the remaining biases together could have produced or hidden the result.

Why restrict the final analysis to first pregnancies?

Restriction removes parity and prior preterm birth, which may have been caused by night work in an earlier pregnancy and so cannot be safely adjusted for.

How long is a PHLT 8520 final methods paper?

Length varies by section, from about five pages to a dozen. The sample is a compact version of about six pages with two tables.