Danish Medical Bulletin    |    Contact Front page  

 
Medication use among acutely hospitalised medical patients
Patient¿s self-reported use versus pharmacy records and drug analyses


Bente Glintborg  


  Download file   (1451 KB)

Accepted by: Health Sciences Copenhagen
Defended on: January 18, 2008
Official opponents: Christian Torp-Pedersen , Birgitte Brock , Jesper Hallas
Tutors: Kim Dalhoff , Henrik Enghusen Poulsen

Published in the PhD Database: January 23, 2008


English abstract
A major reason for medication errors and adverse drug effects is lack of knowledge of the patient¿s in home medication use. This study was carried out at Department of Clinical Pharmacology, Rigshospitalet, Copenhagen. The aim was to describe medication interviews, pharmacy records (PR) and drug analysis as methods to obtain a reliable medication history. Furthermore, to identify if some patient characteristics were predictive of poor self-reporting and if use of some drug types were reported poorer than others.
500 patients admitted to an acute medical department were interviewed about their medication use. Home interviews were performed 1 month after discharge among the patients reporting use amlodipin, bendroflumethiazide, digoxin, glimepiride or simvastatin (the five drugs). The patients provided blood and urine samples. The patients¿ self-reported medication use was compared to PR available from www.medicinprofilen.dk. Blood samples were analysed for contents of the five drugs. A toxicology screening was performed in a subset of 100 urine samples. All analysis results were compared to patients¿ self-reported use.
Included patients had median age 72 years and 60% were women. Median drug use upon admission was 3 prescription only medications (POM). When comparing data from PR vs. self-reported drug use, 19% (95% CI: 15-23%) of POM purchased 0-7 days before admission and 27% (24-29%) of POM purchased 0-30 days before admission was unreported during the interview. Congruence between PR and self-reports was highest for cardiovascular drugs (p<0.05) whereas dermatologicals frequently were underreported. Younger patients reported with of higher congruence. The congruence between interviews and PR was higher upon the home-visits (p<0.05), but still 11% (6-18%) of POM purchased during the preceding week and 18% (15-22%) purchased the preceding month was unreported.
The patients¿ self-reported use of the five drugs was in high agreement with medication analysis (all Kappa>0.79, p<0.05). The 36 patients (7%, 5-10%) that reported in disagreement with analysis results had similar age and sex distribution as patients with congruent data.
Few patients used illicit drugs. Among 12 patients (12%, 6-20%), the toxicology screening was positive for drugs not reported during the interview (cannabinoids: 5 patients, benzodiazepine: 7 patients). The predictive value of self-reported non-use of a drug was high (all negative predictive values >92%).
Despite interviews solely focusing on medication use, patients underreport their medication use. PR might be useful in order to detect errors in the medication history upon hospital admissions ¿ especially among the elderly. Drug analysis and toxicology screens are resource demanding and must be limited to complicated cases.



Danish abstract
En ufuldstændig medicinanamnese giver risiko for medicineringsfejl og mistolkning af symptomer. Formålet med afhandlingen var at vurdere nøjagtigheden af patienters selvrapporterede lægemiddelforbrug når data fra et struktureret interview blev sammenlignet med receptdata og medikamentanalyser. Afhandlingen udgår fra Klinisk Farmakologisk afdeling, Rigshospitalet, og patient-inklusionen foregik på akut medicinsk modtageafsnit, Bispebjerg Hospital.
Interviews blev foretaget blandt 500 akut indlagte medicinske patienter ved indlæggelsen og igen efter udskrivelsen i patientens eget hjem. Receptdata (PR) og medicinoplysninger fra hospitalsjournalen blev indhentet på alle patienter. Blod- og urinprøver indsamlet i relation til interviewene blev analyseret for udvalgte lægemidler (amlodipin, bendroflumethiazid, digoxin, glimepirid, simvastatin) og rusmidler (amfetamin, barbiturat, benzodiazepiner, cannabis, kokain, metadon, opiat), og resultaterne sammenlignet med patienternes selvrapporterede forbrug.
Ved indlæggelsen rapporterede patienterne et median forbrug på 3 receptpligtige lægemidler (POM), hvoraf 6% ikke var registrerede i patienternes PR. Sammenlagt undlod patienterne at rapportere 19% af de POM, som ifølge deres receptdata var indkøbt i ugen op til indlæggelsen. Ved hjemmebesøget var det tilsvarende antal 11%. Hjertemedicin (ATC gruppe C) var den medicintype, som blev rapporteret i højest overensstemmelse med PR. Specielt ældre patienter rapporterede i uoverensstemmelse med receptdata. Overensstemmelsen mellem lægemiddelanalyser og selvrapporteret forbrug af de 5 lægemidler var generelt høj (kappa for alle lægemidler >0,79). En screening for urinindhold af rusmidler blev foretaget blandt 100 tilfældigt udvalgte patienter ved indlæggelsen. Ingen patienter havde positiv screening for amfetamin eller kokain. I alt 12 patienter (12%) havde en screening positiv for stoffer, som ikke var blevet rapporteret ved interviewet (cannabis: 5 patienter, benzodiazepin: 7 patienter). Patienterne var generelt troværdige, når de benægtede brug af de enkelte stoffer (negativ prædiktiv værdi >92%).
Receptdata tyder således på, at der på trods af strukturerede lægemiddelinterviews sker en betydelig underrapportering af lægemiddelforbruget - selv når interviewene foretages i patientens hjem. Dette taler for anvendelse af receptdata i klinikken, specielt blandt ældre patienter. For behandlende læger er patienternes receptdata umiddelbart tilgængelige via www.medicinprofilen.dk. Lægemidler med misbrugspotentiale og illegale stoffer anvendes kun af få patienter. Imidlertid underrapporteres brugen af benzodiazepin og cannabis ofte, og rusmiddel-analyse kan eventuelt overvejes ved diagnostisk uafklarede patienter.