Example: clinical guideline study

This worked example follows a modified e-Delphi study that defines core competencies for rural telehealth nursing, from Setup through two rating rounds to the final report. The study and its numbers are illustrative. S07 is the only rating list on this page. The other round totals are this illustration's counts.

Research question

What competencies must registered nurses demonstrate to deliver safe, effective telehealth in rural primary care settings?

Study configuration

Setting	Value
Workflow	Rate existing items
Method framework	Modified Delphi
Panel target	18 nurses (rural practice ≥ 2 years)
Statements	22 competency statements across 4 areas
Scale	1–9 (the built-in default)
Rounds	2 planned rating rounds
Decision rules	Retain at ≥ 70% rating 7–9; exclude at ≥ 70% rating 1–3; IQR ≤ 2 for either; at least 7 numeric ratings

The 70 percent shares, an interquartile range of at most 2, and a minimum of 7 numeric ratings are this study's prespecified rule. They are also the figures a new study starts with. They are seeds. They are not a standard this page recommends, and this page does not say whether a study should adopt them. A new study starts with three planned rounds. This team planned two.

Rate existing items carries a Recommended pill on the workflow card. That pill is the workflow. It is not a recommendation of these figures.

Areas

Area	Statements	Example statement
Clinical assessment	6	Conducts video-based vital sign interpretation with patient coaching
Technology	5	Troubleshoots connectivity issues without abandoning the visit
Communication	6	Documents telehealth consent before each encounter
Equity & access	5	Adapts teaching for low health-literacy patients via video

Phase 1 — Setup (Week 1)

The same tasks in words. These dates are the illustration's dates, not a product schedule.

• Study design (identity, ethics, method) starts on 2026-01-06 and lasts 2 days.
• Panel design comes after Study design (identity, ethics, method) and lasts 1 day.
• Item bank + Rating & consensus comes after Panel design and lasts 2 days.
• Rounds & feedback comes after Item bank + Rating & consensus and lasts 1 day.
• Review & launch comes after Rounds & feedback and lasts 1 day.

Setup highlights:

• The exemption is recorded on Ethics and data governance as Exempt from review.
• Panel design adds this study's profile fields: years in rural practice, primary setting, and state. A template field is named Years of experience.
• The 22 statements are imported from a CSV. The file needs a column named Statement. This file also has an Area column. Import does not read an ID column. The Item bank assigns the next codes. In this example they run from S01 to S22.
• Rating & consensus asks for a comment when the rating is at or below 6. The control is labelled Low means at or below. On this 1–9 scale, 6 is the top of the uncertain zone.
• Each round runs 14 days, with reminders 7, 3, and 1 days before it closes.
• When the wording is unchanged, Participant feedback can show the median, Middle half of ratings, the zone shares including In between, the person's own earlier rating, and comments a researcher has approved. It does not show a classification, and it does not draw a histogram. The introduction still defines the word Status. There is no Status row. Show a histogram is a Round design switch, and the setup preview can draw bars. That is a different screen.

Stopping advice does not add a round. The principal investigator can add one further rating round as a recorded amendment. What can change after launch is explained on the governed study record.

Phase 2 — Round 1 (Weeks 2–3)

Measure	Result
Invited	18
Enrolled after consent	16
Round 1 submitted	15 of 16 (94%)
Retain	9 statements
Uncertain	9 statements
Polarized	2 statements
Exclude	2 statements

The four classes sum to 22. The participation rate rounds 15 of 16 to 94 percent. That rate is not the stopping test.

Uncertain example (S07):

"Interprets a 12-lead ECG independently during telehealth visits."

Ratings from 15 panelists: 2, 3, 4, 4, 5, 5, 5, 6, 6, 7, 7, 7, 8, 8, 9.

Measure	Value
Median	6
Q1 / Q3	4.5 / 7
IQR	2.5
Rated 7–9	40% (6 of 15)
Rated 1–3	13% (2 of 15)
Class	Uncertain

Neither retain condition is met: 6 of 15 ratings are in 7–9, and the interquartile range of 2.5 is above 2. Exclude also fails: 2 of 15 are in 1–3. The statement is not Polarized. That check needs at least 30 percent in each extreme and at most 20 percent in the middle, from at least 6 ratings. Seven of the 15 ratings are in 4–6, and the low tail is under 30 percent. The class uses those counts. The panelist block rounds 2 of 15 to 13 percent and 7 of 15 to 47 percent. Staff Results also prints 13 percent for the low tail. It has no middle-share column.

Phase 3 — Between rounds (Week 4)

On the Decision stage, a class is not the next step. While another round exists, the 9 Uncertain statements and the 2 Polarized statements are one outcome.

Plan tile	Disposition	Statements
Retained	Hold	9
Excluded	Hold	2
Rated again	Advance	11

The 9 retained statements and the 2 excluded statements are Hold. They are not rated again. The 11 are the 9 Uncertain statements and the 2 Polarized statements, with the wording unchanged. Hold means settled. Advance and Rated again are two labels for the same next step. They are not a reword. Reword it is a separate clarity choice. It leads to To reword and Revise. This example does not use it.

The team confirmed the plan. Confirmation assigns the 11 Rated again statements and does not write new wording. Round 2 contains those 11 statements, still in their Round 1 wording. Because the wording is unchanged, Round 2 feedback shows the previous round's current results, not a snapshot from when the round opened.

Phase 4 — Round 2 (Weeks 5–6)

Measure	Result
Submitted	14 of 16 (88%)
Retain	8 of 11
Uncertain	3 of 11

Round 2 was the last planned round, so those three statements are not given Advance again. The participation rate rounds 14 of 16 to 88 percent. That rate is not the stopping test.

With the Round 1 results, 19 of 22 statements were retained or excluded. The stopping display floors 19 of 22 to 86 percent. The 90 percent seed does not recommend stopping, and that advice does not move a disposition or add a round. The principal investigator could have added one further rating round as a recorded amendment. The team chose not to. The three statements that would have been rated again, and that did not come from a clarity review, are Final: no consensus.

Final results

Final result	Statements
Final: retained	17
Final: excluded	2
Final: no consensus	3

Files produced

The files a manuscript uses are listed on Reporting.

File	Use
Consensus CSV	Results and final results, by the Item bank code
Participation CSV	Participation counts for each round
Responses CSV	One row per answer. The panelist column is a study pseudonym. The file records whether a comment was given, not the comment text.
Study summary (.md)	Counts-only overview

This page does not claim that every export omits a name. An identifiable roster export, and the results-request list, can contain a name and an email. Identity and pseudonyms separates those downloads from the study-report files.

Methods paragraph (excerpt)

We conducted a two-round modified e-Delphi study with 16 rural telehealth nurses. Consensus was defined a priori on a 1–9 scale: statements were retained when at least 70% of ratings were 7–9 and excluded when at least 70% were 1–3, in both cases with an interquartile range of at most 2 and at least 7 numeric ratings. Between rounds, where the wording was unchanged, panelists could see the group median, the middle half of the ratings, the zone shares, their own earlier rating, and comments a researcher had approved. They did not see a histogram or the staff classification. The team did not add a further round.

Lessons learned

Lesson	Where it is recorded
Ask for a comment at or below 6 on this 1–9 scale	Rating & consensus
Rate unsettled statements again as written	Advance and Rated again
Report statements that do not reach consensus	Final: no consensus
Keep study-report files apart from identifiable exports	Reporting, and Identity and pseudonyms

Next steps

• Multi-round worked example — numeric detail
• New study
• Modified e-Delphi