NR-586NP Week 3 Surveillance Data and Descriptive Epidemiology: How to Write It

What NR-586NP Week 3 asks for

Surveillance is the ongoing collection, analysis and return of health data to the people who can act on it, and its forms differ in what they cost and what they miss. Passive systems wait for clinicians to report and undercount reliably. Active systems go looking and cost more. Sentinel systems watch selected sites closely and trade coverage for detail. Syndromic systems read symptoms before diagnoses exist and buy speed at the price of specificity.

Descriptive epidemiology is the discipline of saying what is there before saying why. Person covers age, sex, occupation and other traits that group cases. Place covers where cases live, work or were exposed, and mapping often exposes a cluster no table shows. Time covers trends across years, seasonal patterns, and the shape of an epidemic curve, whose slope hints at a point source or ongoing transmission.

Deliverables around the middle of an 8-week session run 900 to 1,300 words and often ask for a figure: a curve, a table by age group, or a simple map description. If your section runs a discussion this week, expect a data source to be assigned for critique rather than praise, since the useful skill is knowing what a system cannot see.

The NR-586NP Week 3 method, step by step

Six moves turn a raw data source into a descriptive analysis.

1. Write the case definition first

   Clinical criteria, laboratory confirmation and the person, place and time restrictions. Say whether you are counting confirmed, probable and suspected cases, since a definition that shifts mid-analysis creates a trend that is not real.

2. Identify the surveillance system behind your numbers

   Passive, active, sentinel or survey. Name it, then name what it systematically misses: people who never sought care, conditions clinicians forget to report, groups the survey never sampled.

3. Describe by person before anything else

   Break cases down by age group, sex and any occupational or behavioural grouping the data supports. Rates by subgroup, not counts, or the largest group will look like the sickest one.

4. Put the cases on a map, at least in words

   Where do cases cluster, and does the cluster follow a water system, a workplace, a housing type or a boundary in the data. Say what the map cannot separate, since where people live is not always where they were exposed.

5. Plot the cases over time

   Build the curve with an interval suited to the incubation period, then read its shape. A sharp rise and fall suggests a common exposure. A long ragged plateau suggests transmission continuing between people.

6. Convert description into the next question

   Descriptive work ends by naming a hypothesis worth testing, not by declaring a cause. Say which group, place or period deserves the analytic study, and why the description points there.

Sections of a descriptive epidemiology report

Our sizing for an 1,100 word report. Figures do not spend the budget, so build the curve and the table and use the words to interpret them.

Handling surveillance data honestly

Report the case definition with the count. A change in definition can double a case count overnight without a single new infection. Any trend claim needs a sentence confirming the definition held steady.

Undercount is the normal state. Passive reporting captures a fraction of true cases, and the fraction varies by condition and by clinician. Say so once, plainly, and your later interpretation stays defensible.

Distinguish reporting date from onset date. Curves built on the day a report arrived describe an office workflow. Curves built on symptom onset describe an outbreak.

Small numbers in small areas are unstable. A county with four cases can leap to the top of a rate table on one additional case. Suppress or flag rates built on very small counts rather than presenting them as findings.

Describe the map without over-reading it. Clustering by residence can reflect where people sleep, where housing is affordable, or where a testing site sits. Name the alternative explanations you cannot rule out.

Cite the dataset with its version and access date. Public dashboards are revised continuously. The figures you quote are the figures on the day you pulled them, and saying which day makes the analysis reproducible.

Five ways a descriptive report goes wrong

• Counts presented by subgroup instead of rates. The largest age group will always have the most cases. Without rates, the description says nothing except which group is biggest.
• A cause announced from descriptive data. Description generates hypotheses. Declaring the source of a pattern without an analytic design is the reasoning error the week exists to prevent.
• An epidemic curve with an arbitrary interval. Interval choice should follow the incubation period. Weekly bars can hide a point source outbreak completely.
• Silence about what the system misses. Every surveillance source has a blind spot. A report that never names it reads as one that never looked for it.
• Dashboard figures quoted with no access date. Numbers move. Without the date you pulled them, nothing in the report can be checked, and reproducibility rows notice.

Before you submit the descriptive report

• The case definition is written out and held constant across the analysis
• The surveillance system is named along with what it systematically misses
• Person, place and time each get their own treatment
• Subgroup comparisons use rates rather than raw counts
• The epidemic curve interval is justified against the incubation period
• The report ends in a testable hypothesis rather than a declared cause

Questions this week produces

Keep going