Sundhedstjenesteforskning Patient-rapporterede oplysninger/outcomes (PRO) en oversigt Mogens Grønvold, professor, overlæge, dr.med. Forskningsenheden, Palliativ Medicinsk afd. Bispebjerg Hospital Afdeling for Sundhedstjenesteforskning, KU mold@sund.ku.dk Sted og dato Dias 1
Enhedens navn Oversigt Begrebet PRO Principper for valg af spørgskemaer Kort om EORTC QLQ-C30 og FACT-G CAT Symptom-epidemiologi Sted og dato Dias 2
3 PRO/PROM PRO = Patient-Rated Outcomes (variablerne) PROM = Patient-Rated Outcome Measures (instrumenterne) Livskvalitet /QL: Daglig tale for Health-related quality of life (HRQL) Systematisk beskrivelse af sygdoms- og behandlingsrelaterede symptomer og problemer baseret på spørgeskemaer, der udfyldes af patienterne Patient-rated outcomes (PRO) lidt bredere 1. HRQL 2. Tilfredshed med sundhedsvæsenet
4 Hvorfor? Endpoints i trials Deskriptivt: For at vide hvordan patienterne har det For at kunne forebygge/behandle problemer For at forbedre klinisk praksis for den individuelle patient
5 Spørgeskemaer Grundprincip: Finde ud af, hvilke emner, der er relevante (skal A og B sammenlignes, skal begge behandlingers konsekvenser belyses) Vælg det videnskabeligt bedste spørgeskema Se, hvad der mangler - og find det enten i moduler/andre sp.sk. Eller lav selv ekstra items
6 Kriterier for valg af skema Den videnskabelige kvalitet: Validitet/reliabilitet Rigtige indhold, egnet til målgruppen Standardiseret (copyright, autoriserede oversættelser anvisninger for analyse) Andre bruger det (sammenlignelighed) Den videnskabelige kvalitet: Validitet/reliabilitet
Enhedens navn Den videnskabelige kvalitet: Validitet/reliabilitet (1) Klassiske ordbrug Validitet multiple definitioner, fx 1. Indholdsvaliditet 2. Begrebsvaliditet 1. Dimensionalitet skalavalidering, psykometri 2. Concurrent/discriminant validity 3. Klinisk validitet (sammenhænge er som forventet) 3. Kriterievaliditet Reliabilitet 1. Gentagelsesreliabilitet 2. Intern konsistens (Cronbach s alpha) Sted og dato Dias 7
Enhedens navn Den videnskabelige kvalitet: Validitet/reliabilitet (2) Beslægtede begreber/kriterier for valg Responsiveness: Evne til at opfange reel ændring over tid kan sammenlignes mellem metoder. Minimally important difference Effect size/interpretation Referencedata Almenbefolkningen Tilsvarende populationer Erfaring med brug i klinisk praksis Egenskaber som screeningsmetode Resultater Grafisk fremvisning Kobling til kliniske retningslinjer/interventioner Sted og dato Dias 8
9 Standardspørgeskemaer EORTC* QLQ-C30 (oftest i Europa) Moduler, ca. 20 fuldt validerede samt ca. 20 delvist validerede FACT-G (oftest i USA) Moduler Begge er multidimensionelle og består af ca. 30 items, men EORTC scores til 15 dimensioner (træthed, smerter, psykisk velbefindende, etc.) FACT-G scores til fire områder eller en samlet, der også medregner modulet. (Endimensionelle mål: Utility (efterspørges af sundhedsøkonomer, men klinisk af begrænset værdi)) *) Interessekonflikt: 24 års medlemskab, utallige forskningssamarbejder, nuværende formandskab for, og bevillingsmodtager fra EORTC Quality of Life Group
Enhedens navn CAT Sted og dato Dias 10
Computer-adaptive testing (CAT) Ask a question Use the response to this question to estimate the level Select next question based on this knowledge Use the response to this question and the previous one to re-estimate the level - and so on... Stop when the desired level of precision is reached FSV kursus i spørgeskemakonstruktion
Advantages of CAT Individually adapted questionnaires with fully comparable scores across individuals Increased precision (smaller samples required) Reduced response burden (and no stupid questions ) Reduced floor and ceiling effects Questionnaire length selected for each study or person Immediate calculation of scores FSV kursus i spørgeskemakonstruktion
EORTC procedure for CAT development Phase 1: Literature review and interviews Phase 2: Develop, select and pilot test new candidate items Phase 3: Data collection (paper or computer administered) and analyses Total N>1000, data from at least three countries, requires centres capable of fast data collection Analyses: remove poor items, estimate IRT model, develop CAT program, pilot testing CAT Phase 4: Test CAT in large samples Implement CAT technology Collect data from 7-10 test centres (2013-2016) FSV kursus i spørgeskemakonstruktion
FSV kursus i spørgeskemakonstruktion Patient 1 Do you have any trouble taking a short walk? Not at all Do you have any trouble taking a long walk? A little Do you have any trouble running 1 km? Quite a bit Completed. PF score = 70
Patient 1 Patient 2 FSV kursus i spørgeskemakonstruktion Do you have any trouble taking a short walk? Do you have any trouble taking a short walk? Not at all Very much Do you have any trouble taking a long walk? Do you need help dressing? A little A little Do you have any trouble running 1 km? Do you need to stay in bed or a chair during the day? Quite a bit Quite a bit Completed. PF score = 70 Completed. PF score = 20
Enhedens navn CAT item banks CAT fx brug af de bedste 5-10 items per emne per patient Udvalgt live Individuelt Elektronisk Short-forms fx brug af de bedste 5-10 items per emne per patient Udvalgt til studiet Ud fra den forventede population Papir eller elektronisk Sted og dato Dias 16
Enhedens navn Mange andre ting betyder lige så meget som skema De tre vigtigste er... Sted og dato Dias 17
Enhedens navn Mange andre ting betyder lige så meget som skema De tre vigtigste er... Komplethed, komplethed og komplethed Sted og dato Dias 18
Enhedens navn Mange andre ting betyder lige så meget som skema De tre vigtigste er... Komplethed, komplethed og komplethed Timing fanger man det rette? Er timingen retfærdig i sammenligning Information, viden, instruktion (personale, patienter) Logistik masser af detaljer Protokollen Alle de sædvanlige ting, lige så vigtige Analyse Præ-specificeret i protokol Oftest komplicerende pga. missing data og evt. multiple outcomes Sted og dato Dias 19
Enhedens navn Symptom-epidemiologi 3LNQ Sted og dato Dias 20
21 Symptom-epidemiologi i DK Nationalt repræsentativ stikprøve af patienter Kræft stadium 3 eller 4 Kontakt med et sygehus i seneste 12 mdr. Fra 54 hospitalsafdelinger i 3 amter (Ringkøbing, Fyn, H:S) Spurgte 1.630 Svar fra 977 (60%)
22 Spørgeskema 1. EORTC QLQ-C30 (30 spørgsmål om 15 emner)
23 EORTC QLQ-C30 symptom hyppigheder. Let Svær grad Fatigue 57 22 Nausea and vomiting 13 4 Pain 39 17 Dyspnoea 37 14 Sleeping difficulties 48 19 Lack of appetite 30 16 Constipation 26 19 Diarrhea 29 8 Financial difficulties 21 9
24 Samlet Symptom/problemhyppighed 4,8 i let grad (mindst lidt) 1,7 i svær grad (mindst en del)
Enhedens navn DSR Sted og dato Dias 25
26 Doctor s recognition The questionnaires thus cover the symptoms patients are treated for by their doctors However, do the medical records cover the symptoms experienced by patients? Aim: to examine the extent to which the initial symptomatology reported by patients in self-assessment questionnaires is covered by their medical records
27 Methods We used the data about contents of medical records 58 patients had completed a questionnaire (the study was started later) We identified the patients having reported each symptom/problem ( A bit or Very much ; score > 34) in the questionnaire (EORTC QLQ-C30, ESAS; HADS) How many of these were identified in the medical records at admission?
Results: Detection rates 96% pain 28
29 Results: Detection rates 96% pain 54% psychol. distress 52% nausea/vomiting 46% dyspnea 44% lack appetite 38% fatigue 38% constipation 36% physical funct. 35% conc./memory 31% diarrhea 27% depression 22% anxiety 15% difficulty sleep 12% drowsiness 8% economic diff. 6% social contact
30 Conclusions Many symptoms were undetected Nurses are not better (except for physical function) These figures must be interpreted carefully: patients may not remember everything, they may explain that a symptom is unimportant, and the doctor may have to prioritize when talking to the patient for the first time, etc. More systematic screening for common problems may increase detection and thus, potentially, treatment Stromgren et al., J Pain Sympt Manage: 2001
31 Klinisk praksis
Fig 1. Example of individual quality-of-life profile presented to physicians in the intervention arm Velikova, G. et al. J Clin Oncol; 22:714-724 2004 Copyright American 32 Society of Clinical Oncology
Fig 2. Flow diagram of the progress through the randomized study Velikova, G. et al. J Clin Oncol; 22:714-724 2004 Copyright American 33 Society of Clinical Oncology
Fig 4. Proportions of patients showing clinically meaningful improvement, no change, or deterioration in Functional Assessment of Cancer-General (FACT-G) score after three encounters, by study arm Velikova, G. et al. J Clin Oncol; 22:714-724 2004 Intervention and A-c different from control Copyright American 34 Society of Clinical Oncology