If roughly the same share of people distrusts health care organizations in two surveys, their individual views could still have changed. In a new analysis of a US panel, the overall distributions moved little between 2021 and 2022 while the model estimated frequent movement between profiles.
- The authors analyzed 1,437 adults who took part in both survey waves and met the analysis criteria.
- About 65% in 2021 and 67% in 2022 fell into the two higher mistrust profiles, using the study’s rounded model estimates.
- Among people starting in the moderate profile, the estimated chance of remaining there was 35%. The profile and transition are both model estimates.
- Paper date
- 20 November 2026
- Analytic sample
- 1,437 US adults
- Survey waves
- April–May 2021 and 2022
Original publication: 20 Nov 2026 · The date above refers to this brief.
Estimated chance of staying in the same profile
percent within each 2021 starting profile
How a panel reveals a change
Measure twice
Ask the same people about their views at two points in time.
Estimate profiles
Group patterns of answers using a statistical model.
Track transitions
Estimate moves between profiles even when overall shares are similar.
The same people answered twice
Allen and colleagues used a US probability panel surveyed in late April and early May of 2021 and 2022. Of the 1,449 adults the Methods say completed both surveys, 12 had incomplete demographic responses, leaving 1,437 for analysis. The questions covered perceptions such as privacy, competency and whether organizations could be misleading. The unit of measurement was a respondent’s view of health care organizations during those years.
Source 1 ↗What the headline totals miss
The authors used a statistical model to group response patterns into four profiles. The estimated moderate and strong profiles accounted for about 65% together in 2021 and 67% in 2022. Imagine 100 people in a separate, hypothetical survey. If 10 leave a category and 10 enter, the category total stays the same even though 20 people switch. Repeated measurements on individuals give researchers a way to investigate that hidden movement.
Source 1 ↗How to read the 35% figure
The estimated probability of staying in the moderate profile was 0.35 for respondents who started in that profile. For people who started in the strong profile, the corresponding estimate was 0.26. These probabilities are conditional on the starting profile, while the 65% and 67% figures summarize the whole weighted sample. The model incorporated uncertainty in class membership. The authors caution that some apparent switching may reflect imperfect classification.
Source 1 ↗What a later study could find out
Closer survey intervals could show whether a change persisted or reversed before the next annual wave. Asking separately about hospitals, insurers and public health agencies could also reveal where views changed. Those are follow-up questions prompted by this study. The two waves span about a year and the estimates describe that historical period.
Source 1 ↗Read the starting group before reading the probability
Primary journal article. Checked Methods, Results, Table 2, Limitations and funding and conflict disclosures. Publisher metadata and abstract also checked. Supplements, source-data files, code and clinical records were not independently audited. No publisher figure reproduced.
01Can stable totals coexist with individual change?
- What was observed
- Weighted estimates of the four profiles were similar in two surveys. The transition model estimated considerable movement within the panel.
- Where the conclusion stops
- The estimated share in a profile at one date and the probability of staying in it answer different questions.
Source 1 · Journal Results and Table 2
02What do the four labels describe?
- What was observed
- The model identifies recurring patterns of answers to seven questions about health care organizations.
- Where the conclusion stops
- The model can assign some responses to a different profile even when an underlying view has changed little.
Source 1 · Journal Methods and Limitations
Numbers you can inspect
| Measure | Value & unit | Origin & method |
|---|---|---|
| Final analytic sample | 1437 people | Reported Adults with eligible data from both survey waves. Source 1 · Journal Methods and Results |
| Moderate or strong mistrust, 2021 | 65 percent | Calculated (0.33 + 0.32) × 100, using rounded modeled shares. Source 1 · Journal Table 2 |
| Moderate or strong mistrust, 2022 | 67 percent | Calculated (0.34 + 0.33) × 100, using rounded modeled shares. Source 1 · Journal Table 2 |
| Retention in no-mistrust profile | 0.22 probability | Reported Conditional on starting in the no-mistrust profile. Source 1 · Journal Table 2 |
| Retention in neutral profile | 0.18 probability | Reported Conditional on starting in the neutral profile. Source 1 · Journal Table 2 |
| Retention in moderate profile | 0.35 probability | Reported Conditional on starting in the moderate profile. Source 1 · Journal Table 2 |
| Retention in strong profile | 0.26 probability | Reported Conditional on starting in the strong profile. Source 1 · Journal Table 2 |
Compare the actual experiments
These studies answer different questions. Read the unit and endpoint before comparing results.
| Study | Unit & setting | Readout | Interpretation boundary |
|---|---|---|---|
| Wave-level distribution Source 1 · Journal Table 2 | Weighted probability sample of US adults | Four estimated profile shares per survey wave | An overall distribution provides a snapshot of one date. |
| Within-person transition model Source 1 · Journal Methods, Table 2 and Limitations | Respondents surveyed at both dates | Conditional probabilities of moving or staying | Profile assignment is uncertain and responses were collected about a year apart. |
The export includes claims, available numbers, methods and source locations. It contains our reading notes and published summaries; it is not raw participant data or an independent reanalysis.
Evidence update · 21 Nov 2026
First publication. Reported observations, arithmetic and proposed follow-up questions are separated. No raw-data reanalysis. AI source check; no human editorial or clinical review.
Three numbers with different denominators
| Reading question | What the paper reports | What remains open |
|---|---|---|
| 1,437 | Adults included in the analysis of the two survey waves | The paper uses survey weights to estimate population patterns. |
| About 65% and 67% | Combined moderate and strong profiles in 2021 and 2022, calculated from rounded estimates | These are summaries for two dates. |
| 35% | Estimated chance of remaining in the moderate profile among those who started there | The denominator is the modeled starting profile. |
| 26% | Estimated chance of remaining in the strong profile among those who started there | Uncertainty in profile membership can affect transitions. |
Original Press-News reading framework. Follow-up questions are our interpretation, not experiments performed by the authors.
Your questions, answered
When were people surveyed?
The two waves used for this analysis ran from April 23 to May 3 in 2021 and again in 2022. The research article appeared in 2026.
What was measured?
A seven-item index asked about perceptions of health care organizations, including questions about privacy and competency.
Is 35% a share of all respondents?
It applies to the model’s moderate starting profile. It estimates the probability of belonging to that same profile one year later.
How certain are the individual transitions?
The model incorporates uncertainty in the starting and ending profiles. The authors acknowledge that uncertainty in profile assignment may account for some apparent moves.
Limits of this interpretation
- Self-reported perceptions can depend on question wording and recent experiences.
- The four profiles are inferred from responses, so movement is sensitive to how the model assigns membership.
- An interval of about one year leaves short-term changes unobserved.
- The questions group health care organizations together. Views of specific institutions or practitioners may differ.
Sources & transparency
- Allen, Fu and Cuevas (2026): Mistrust in Health Care Organizations Among US Adults
Primary journal article. Checked Methods, Results, Table 2, Limitations and funding and conflict disclosures. Publisher metadata and abstract also checked. Supplements, source-data files, code and clinical records were not independently audited. No publisher figure reproduced. · Accessed 21 Nov 2026
DOI: 10.1001/jamanetworkopen.2026.43935
Prepared and source-checked with AI. Press-news Team is the collective publication byline, not a medical reviewer. No human editorial or clinical review has taken place. This educational article examines published observational research. Claims and calculations are based on the sources listed below. Photographs provide context for the topic.
Source check: AI source check — selected primary results, captions and study boundaries
Clinical review: Not applicable to this educational guide
Suggest a correction