Methods	RCT, two centres Recruitment: December 1996 to December 1998. Duration of follow up: Six months from recruitment (plus additional three months)
Participants	Country: USA Participants: 102 patients (66 males, 65%) in intervention group, 98 (55 males, 56%) in comparison group. Actual age of study subjects at recruitment: median 63.5 years (range 25-88 years) Male sex: 61% Ethnicity: ‘white’ 64% Actual severity of heart failure in study subjects at baseline: NYHA class, n (%): controls II 33 (34%), III 60 (61%), intervention group II 38 (37%), III 57 (56%). LVEF: control group mean 27.5% (SD 13.9, range 5-60), intervention group 27.1% (SD 13.8, range 10-70) Study inclusion criteria: Admitted to one of two hospitals with a primary diagnosis of NYHA class III/IV CHF. English speaking. Permission from patient’s ‘primary physician’ Judged to be at high risk of CHF re-admission, ie one or more of the following criteria: “Age >70 years. “LVEF <35%. “One or more other hospital admission for CHF in previous year. “Ischaemic cardiomyopathy. “Peripheral oedema at hospital discharge. “Less than 3kg weight loss while in the hospital. “Peripheral vascular disease. Or any one of the following during the index admission: “Pulmonary capillary wedge pressure >25 mmHg. “Cardiac index <2.0 l/min/m2. “SBP >180 mmHg. “DBP >100 mmHg. Study exclusion criteria: Valvular heart disease requiring surgery. Active substance abuse. Cardiomyopathy (peripartum, hypertrophic with LV outflow tract obstruction or restrictive). Constrictive pericarditis. Psychiatric disease. Dementia likely to limit compliance. Non-cardiac illness likely to cause repeat hospital admission. Heart transplantation likely to occur within six months. Uncorrected thyroid disease. Serum creatinine => 3.0 mg/dl. Long term home intravenous inotropic therapy. Cardiac surgery or MI during the index admission. Active participation in another research trial. Residence in a nursing home, rehabilitation facility or outside the area served by the two hospitals
Interventions	Duration of intervention: 6 months. Intervention Group: ’multidisciplinary program’ During index hospitalisation: CHF cardiologist designed an individualised treatment plan for each patient before randomisation which included medication, diet and exercise management After discharge: ‘Telephone nurse co-coordinator’ phoned patients within 72 hours of discharge and then weekly for 1st month, bi-weekly in 2nd month and then monthly. (Content of phone calls: set script with problems pursued as clinically indicated . No medication adjustments over phone.) Monthly follow up with CHF nurses (usually in CHF clinic). ‘Primary care physicians’ (66% internal medicine physicians, 29% cardiologists) received regular updates from CHF nurses and were notified of abnormal lab results. All intervention patients received: pill sorter, list correct medications, list of dietary and exercise recommendations, 24 hour telephone contact number and patient educational material. If required and financial resources limited patients also received: 3g sodium ‘Meals on Wheels’ diet, weigh scale, medications, transport to the clinic and a phone. CHF cardiologist saw patients at 6 months. Content of CHF nurse follow up: aimed to implement the treatment plan designed by CHF cardiologist by using a pre-specified 55 page algorithm (also designed by the CHF cardiologists) which included initiation and titration of drugs, a low sodium diet and exercise recommendations Comparison group: Usual care. This was care by the patients’ primary physicians (73% internal medicine physicians, 26% cardiologists). CHF cardiologist designed treatment plan for each patient “documented in patient’s chart without further intervention”
Outcomes	Primary endpoint: Total number of CHF hospital admissions plus all cause deaths (i.e. composite endpoint) Secondary outcomes: Death. CHF hospital admissions. All cause hospital admissions. Change in HRQOL (MLHFQ). Change in activity status (Duke Activity Status Index). Process indicators including: proportion of patients with systolic dysfunction receiving ACEI according to published guidelines or appropriate alternative treatment if intolerant of ACEI; percentage patients euvolemic according to defined goal weight; compliance with dietary guidelines using locally developed sodium score and cost data
Notes	Data source: published data and information supplied by author* Generalisability: 1,452 patients with heart failure were screened, (screened patients were not consecutive admissions*); 976 (67% of those screened) met inclusion criteria of whom 686 (70%) had one or more exclusion criteria; of the remaining 290 eligible subjects 90* (31%) refused to participate Consort flow chart: not supplied Rationale for sample size: supplied Other points: Reason for exclusions (based on 776 patients, i.e. 686 excluded patients and 90 patients who refused): participating in another research protocol 15.5%; renal dysfunction 14.3%; dementia or substance abuse 10.7%; planned cardiac revascularisation or heart transplant 10.2%; cardiac exclusions such as hypertrophic cardiomyopathy, restrictive cardiomyopathy, amyloidosis, valvular heart disease 10.2%; living outside catchment area 7%; discharged to nursing home or on intravenous inotrope 5.1%; non-cardiac disorder likely to cause repeated hospital admission 12.8%; primary care physician declined to participate 2.3% Intervention and control groups appear similar at baseline but some inclusion criteria data not presented Three patients required heart transplantation (two during study and one immediately after) despite the exclusion criterion of heart transplantation likely to occur within six months Funding: Partial funding was provided by CardioContinuum, Inc. This company, the University and its Division of Cardiology are entitled to royalty on the use of the CHF management programme described in the study. The University also owns CardioContinuum stock. None of the investigators has personal royalty interests, stock or consulting arrangements with CardioContinuum, except for one who was once a CardioContinuum employee. Generation of randomisation sequence and allocation concealment: “The coordinating centre made treatment assignments by using an automated telephone response system…Random number schedules were prepared before initiation of patient recruitment ands were unknown to the clinical investigators.” Comment on care giver performance bias: physicians providing usual care were aware of study and knew that their patient had not been allocated to the intervention so possible Hawthorn effect on care received by the usual care group. *Also 10 % of these physicians had other patients allocated to the intervention arm which may also have influenced their usual care. The effect of both of these influences would be to underestimate the effect of the intervention. Risk of attrition bias: low. Risk of detection bias: low for main outcomes.
Risk of bias
Item	Authors’ judgement	Description
Allocation concealment?	Yes	A - Adequate