Blog · Board and management reporting · Healthcare
Compare aggregate patient survey scores and response rates by location. A worked example shows how changing response mix can move a network score even when location scores stay unchanged.
A patient satisfaction dashboard compares survey scores with response rates and the mix of locations supplying the answers. Show valid response counts alongside every score, because a larger network score can come entirely from a change in who responded. Use aggregate, de-identified survey summaries to support an operational service review.
The central question is how respondents rated a defined aspect of their experience. A satisfaction question asks for an evaluation; an experience question can ask whether something happened. Preserve the original wording and scale so a monthly comparison measures the same thing.
For a top-box measure, count valid answers in the designated most positive category and divide by all valid answers to that question. AHRQ's CAHPS data tools define top-box results this way, with the positive category depending on the item. A single highest category and a grouped highest-rating range are different definitions, so label whichever your survey uses.
Top-box score = Most positive valid answers / All valid answers to the question
Invitation response rate = Completed surveys / Eligible invited survey units
Location response share = Location's valid answers / Network's valid answers
Here, a survey unit is one eligible invitation under a consistent invitation policy. Repeated visits and follow-up reminders need a defined treatment before the denominator is created. The invitation response rate is an internal operating measure; use the applicable survey program's prescribed calculation for official reporting.
Ask the survey administrator for one row per period, location, question, and survey version. Include eligible invitations, completed surveys, valid item answers, and counts in each response category. Keep the completion rule and response cutoff beside the file.
Invitations and survey completions can repeat across question rows. Calculate participation from a separate period-and-location summary rather than summing those repeated values across questions. Use item answers only for the relevant question's score.
Leave out names, contact details, visit identifiers, medical information, and free-text comments. Apply your organization's minimum group size and suppression rules before sharing a summary. Check combinations of locations and subgroups too: a hidden small cell can sometimes be inferred by subtracting visible cells from a total.
The survey dashboard guide covers general survey structure. An employee satisfaction dashboard uses related denominator checks for workforce surveys, with a different reporting population.
Consider an invented US outpatient network with East and West locations. Its internal question asks about overall satisfaction, and the highest response category is the top box. This example is a local management measure, not an official HCAHPS score.
Every completed survey in this example contains a valid answer to that question. In a real file, completed surveys and valid answers can differ. June 2026 results are:
| Location | Eligible invitations | Completed surveys | Valid answers | Top-box answers | Other valid answers |
|---|---|---|---|---|---|
| East | 300 | 180 | 180 | 126 | 54 |
| West | 200 | 80 | 80 | 64 | 16 |
| Network | 500 | 260 | 260 | 190 | 70 |
Each result below follows from those counts. Display percentages to one decimal place, rounding only after division.
| Location | Response rate calculation | Response rate | Top-box calculation | Top-box score |
|---|---|---|---|---|
| East | 180 / 300 | 60.0% | 126 / 180 | 70.0% |
| West | 80 / 200 | 40.0% | 64 / 80 | 80.0% |
| Network | 260 / 500 | 52.0% | 190 / 260 | 73.1% |
West's score is higher among respondents, but its participation rate is lower. That calls for checking the invitation process and response coverage before drawing a conclusion about the locations' service. The network score weights answers, so an equal average of location percentages would answer a different question.
Now change the response distribution while keeping invitations and each location's top-box rate unchanged. July has the following counts:
| Location | Eligible invitations | Completed surveys | Valid answers | Top-box answers | Other valid answers |
|---|---|---|---|---|---|
| East | 300 | 100 | 100 | 70 | 30 |
| West | 200 | 160 | 160 | 128 | 32 |
| Network | 500 | 260 | 260 | 198 | 62 |
East remains at 70 / 100 = 70.0%, and West remains at 128 / 160 = 80.0%. Network participation remains 260 / 500 = 52.0%. Yet the network score rises to 198 / 260 = 76.2%, up 3.1 percentage points from June using unrounded calculations.
The higher-scoring location now supplies more of the valid answers. For a supplementary comparison, hold location weights at their invitation shares: East 300 / 500 = 60.0% and West 200 / 500 = 40.0%. Applying those fixed weights to either month's location scores gives 0.60 × 70.0% + 0.40 × 80.0% = 74.0%.
Label that result as a fixed-location-weight comparison. It does not correct nonresponse within a location or reproduce formal patient-mix adjustment. CMS explains that publicly reported HCAHPS results require adjustments for survey mode and patient characteristics; use that program's approved methods and outputs for its official comparisons.
For June, 126 + 54 = 180 and 64 + 16 = 80. Across the network, 190 + 70 = 260 valid answers. July similarly reconciles: 70 + 30 = 100, 128 + 32 = 160, and 198 + 62 = 260. These identities catch missing or duplicated response categories.
Tie invitations and completions to the survey administrator's control report. Use the business spreadsheet data quality checks to review blanks and duplicates before comparing periods. If item answers are missing, show that count separately; do not force the response categories to equal all completed surveys by treating blanks as dissatisfied answers.
The patient experience lead owns the interpretation and action. The survey administrator checks whether invitation eligibility, reminders, languages, or response windows changed. Location managers investigate service processes after those checks. The analyst maintains the denominators and comparison basis. None of these aggregate scores establishes a clinical outcome or proves that an intervention caused a change.
Bring an approved aggregate survey file or Google Sheet; Covirage maps the columns, and its tools compute and check scores, denominators, and period comparisons. The external AI model explains the results with every figure traced to its rows. Create a free Covirage account and bring a sample of your own de-identified summary.
See self-service analytics with Covirage for the supporting approach.
It is the share of valid answers in the most positive response category defined for that question. Identify the scale and category explicitly. Do not combine questions with different scoring rules or treat unanswered questions as negative answers.
The score describes people who answered the question. Response rate describes participation in the survey invitation population. A change in participation can change who is represented, so read both measures before attributing a score movement to a service change.
A simple internal calculation does not reproduce official results. HCAHPS uses specified administration and reporting methods, including adjustments. Keep locally calculated scores clearly labeled, and use the approved official figures when reporting the program's published measures.
The patient experience lead should own question definitions, survey periods, and follow-up actions. The survey administrator confirms invitation and response counts. An analyst checks calculations, while the organization's privacy owner approves aggregation and suppression before results are shared.