Table 1.
Risk of bias assessments of RCTs.
| Type of risk of bias | Risk of bias: authors' judgement | Support for judgement |
|---|---|---|
| Aiken et al. (2006) [27] | ||
| Random sequence generation (selection bias) | LOW | “Randomization was carried out within diagnosis, in blocks of 30 patients (15 Phoenix Care, 15 control) by a member of the project administration staff. Sealed envelopes, color-coded by diagnosis and containing the assignment to condition, were shuffled and assigned to participants in order of shuffling.” |
| Allocation concealment (selection bias) | LOW | “Randomization was carried out within diagnosis, in blocks of 30 patients (15 Phoenix Care, 15 control) by a member of the project administration staff. Sealed envelopes, color-coded by diagnosis and containing the assignment to condition, were shuffled and assigned to participants in order of shuffling.” |
| Blinding of outcome assessment (detection bias) | LOW | “Every 3 months all participants received a 30- to 45-minute telephone interview by a professional interviewing firm; interviewers were blind to condition and diagnosis.” |
| Incomplete outcome data (attrition bias) | LOW | “At the end of data collection 44% of the PhoenixCare participants and 25% of the control patients were still participating…percentages for PhoenixCare versus controls, respectively were 16% versus 13%, death; 12% versus 13%, hospice; …6% of PhoenixCare declined and 11% controls declined to continue participation, another 10% and 14% respectively, disqualified by leaving their MCO. [Managed Care Organization] “Only one condition by retention interaction was detected that signaled differential attrition, that for having been given sufficient information and education to manage illness at home, P < 0.05.” |
| Selective reporting (reporting bias) | LOW | No selective reporting |
|
| ||
| Long and Marshall (1999) [37] | ||
| Random sequence generation (selection bias) | UNCLEAR | “For a randomized trial of ambulatory case management, 317 enrollees in the Kaiser Permanente Medical Care Program, Ohio who were 75 years and over, had severe functional disability, or had excessive hospital or emergency department (ED) use were randomly assigned to a Regular Care Group or a Case Managed Group.” |
| Allocation concealment (selection bias) | UNCLEAR | No statement |
| Blinding of outcome assessment (detection bias) | HIGH | “Case managers became integral members of the care team, which included the client's personal physician and the physician advisor, who developed the initial care plan for each client. The case managers were responsible for making periodic home visits, reporting back to the care team, and revising care plans as necessary. While case managers made at least one home visit every 6 months, weekly visits to some clients were not uncommon. In addition to this, the case manager scheduled medical appointments, accompanied patients on these appointments and arranged for nonmedical services such as respite care, meals-on-wheels, nursing home placement, Medicaid eligibility, and transport to and from the physician.” |
| Incomplete outcome data (attrition bias) | LOW | This was a study of care in the last month of life. “the two groups of deceased are statistically comparable to such an extent as to suggest that statistical benefits of the initial random assignment persisted even in death.” |
| Selective reporting (reporting bias) | LOW | No selective reporting |
| Other bias | LOW | No other biases ascertained |
|
| ||
| Meier et al. (2004) [38] | ||
| Random sequence generation (selection bias) | UNCLEAR | “Care Coordinator nurses were randomly assigned to provide either usual case management (4 nurses) or the palliative care enhanced intervention (5 nurses).” |
| Allocation concealment (selection bias) | UNCLEAR | No statement |
| Blinding of outcome assessment (detection bias) | UNCLEAR | No statement |
| Incomplete outcome data (attrition bias) | NOT APPLICABLE | Program description, no quantitative results |
| Selective reporting (reporting bias) | NOT APPLICABLE | Program description, no quantitative results |
| Other bias | NOT APPLICABLE | Program description, no quantitative results |
|
| ||
| Naylor et al. (1999) [28] | ||
| Random sequence generation (selection bias) | LOW | “assigned to study group using a computer generated algorithm” |
| Allocation concealment (selection bias) | LOW | “RAs, who were responsible for enrolling patients in the study were blinded to study groups and hypotheses” |
| Blinding of outcome assessment (detection bias) | LOW | “Outcome data were collected by RAs blinded to study groups and hypotheses” |
| Incomplete outcome data (attrition bias) | LOW | Attrition rate (including deaths) in intervention group 30%, in control group 26%. “For patients who did not complete the entire 24-week postindex hospitalization study period (death or withdrawal), data collected between randomization and withdrawal were used in the analyses, performed according to the intention-to-treat principle,” “The 262 patients who completed the study and the 101 persons in the attrition group did not significantly differ in sociodemographic variables and severity of illness measures (e.g., number of comorbid conditions).” |
| Selective reporting (reporting bias) | LOW | No selective reporting |
| Other bias | LOW | No other biases ascertained |
|
| ||
| Nickel et al. (1996) [39] | ||
| Random sequence generation (selection bias) | UNCLEAR | “Participants were stratified by agency and randomly assigned to the case-managed or usual-care groups in precoded blocks of two” |
| Allocation concealment (selection bias) | UNCLEAR | No statement |
| Blinding of outcome assessment (detection bias) | HIGH | “The schedule of data collection included administration of the ADL and IADL scales by agency nurses at intervals of at least 1week and monthly administration of the QWB by the data collector.” “Since nurses at the seven participating agencies could be assigned to both experimental and control patients, diffusion of intervention practices was of concern.” “Both the case managed and usual-care groups received monthly home visits by project staff for assessment of quality of life outcomes.” |
| Incomplete outcome data (attrition bias) | HIGH | “Scores for deceased subjects were entered as zero at the monthly time points following occurrence of death. Missing QWB scores in living patients were imputed through linear regression with predicted scores based on individual-specific ADL and IADL scores at time points proximal to the missing QWB times. For time points with ADL/IADL scores also missing, values were imputed through maximum likelihood estimates incorporated within the BMDP program”. |
| Selective reporting (reporting bias) | LOW | No selective reporting |
| Other bias | LOW | No other biases ascertained |
Risk of bias was assessed according to the methods recommended in: Higgins JPT, Green S (editors). Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane Collaboration, 2011. Available from http://handbook.cochrane.org/ [22].