Triller 2007.

Study characteristics
Methods	RT Study duration: 201 days (3‐week intervention and 180 days of follow‐up)
Participants	154 participants Setting: home Diagnostic criteria: participants had to have a primary or secondary diagnosis of heart failure and were referred to receive skilled nursing services Age (years) (mean ± SD): control: 78.1 (11.2), intervention: 81.3 (9.3), participants had to be ≥ 21 years Female n (%): control: 55 (72), intervention: 56 (73) Country: USA Comorbidity: heart failure Sociodemographics: the catchment area provided participants from urban, suburban, and rural environs and from across all socioeconomic classes. According to census data, 89% of the population of these 3 counties combined is white, and 87% of adults have a high school diploma. Median household income for the counties is approximately USD 47,000 (2003 data). Non‐English‐speaking participants were included if adequate translation services were available from family members or friends. Ethnicity: unknown; it is difficult to ascertain the ethnicity from the information given. control 68 (88%), intervention 75 (97%) Date of study: 1 July 2002 to end 2004
Interventions	1 intervention: pharmaceutical care services Pharmaceutical care services consisted of an initial comprehensive in‐home medication assessment (concurrent with agency admission) and 2 follow‐up visits (7‐10 and 18‐21 days later). The follow‐up visits were contingent on the participant’s continued receipt of visiting nurse services (i.e. participants discharged from the visiting nurse before 21 days would not receive all of the pharmacist’s planned visits). Throughout the 3‐week intervention period, the clinical pharmacist accessed and reviewed all pertinent physician notes and laboratory test values via the National Endowment for the Humanities data system and interacted with prescribers on behalf of the participants as necessary. Control participants received the usual care provided by the visiting nurse association. Visiting nurse services (provided to both groups) included basic nursing care and a brief physical assessment and medical history.
Outcomes	Hospitalisations and mortality were assessed during a 180 day follow‐up period.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	This was adequate: "Patients provided informed consent for study participation and were randomised to receive usual care or usual care plus pharmaceutical care by means of a computer‐generated random numbers table in blocks of four."
Allocation concealment (selection bias)	Low risk	This was adequate: "Once informed consent was received, the nurse obtained a baseline quality‐of life assessment (using the SF‐12) and then accessed a sealed envelope containing the group assignment from the intake office."
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Blinding or a lack of blinding was unlikely to affect the outcome because the usual care group received usual care from nurses whereas the intervention group received usual care plus the services of a pharmacist
Incomplete outcome data (attrition bias) All outcomes	Low risk	Attrition in both the intervention and control groups was comparable
Blinding of outcome assessment (detection bias) All outcomes	Low risk	This is of less concern in our case as all of our outcomes are objective (according to the EPOC risk of bias criteria).
Selective reporting (reporting bias)	Low risk	No evidence of a problem
Protection against contamination bias	Low risk	There is no evidence of contamination between the intervention and the control groups
Other bias	High risk	Small sample may have produced low power to detect an effect. Poor pharmacist‐prescriber communication may have reduced efficacy of intervention.