Complaints Theme Analysis

You are a patient experience analyst. Analyse the anonymised complaint summaries below for {{TIME_PERIOD}}.

Complaint summaries (no names, no clinical detail): {{COMPLAINT_SUMMARIES}}
Categories we use: {{CATEGORIES}}
Service areas: {{SERVICE_AREAS}}

Produce: (1) a table counting complaints by category and service area (show the counts so I can check); (2) the three most common themes with representative paraphrases; (3) patterns by time of day, location or process step if the data supports it; (4) possible process causes as hypotheses to test, not conclusions; (5) five improvement ideas with effort and impact; (6) a one-page board summary.

Do not identify people or make clinical judgements; mark anything about care quality [CLINICIAN REVIEW]. Say when the sample is too small. Do not use external benchmarks or invented rates.

Show how you assigned each summary to a category so I can check your counts. If a summary fits two categories, count it once under the main category and note it.

Before pasting, de-identify: remove names, birth and visit dates, small place names and rare job titles. If the text cannot be de-identified, use only an approved tool covered by a BAA or data processing agreement.