بررسی نقش یادآورهای کامپیوتری عرضه شده بر کاغذ برای متخصصان مراقبت سلامت: تاثیرات بر عملکرد حرفهای و پیامدهای مراقبت سلامت
Referencias
منابع مطالعات واردشده در این مرور
منابع مطالعات خارجشده از این مرور
منابع اضافی
منابع دیگر نسخههای منتشرشده این مرور
Characteristics of studies
Characteristics of included studies [ordered by study ID]
Ir a:
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: blood pressure ≥ 100, "poor" follow‐up, women: 49%, mean age: 43, n = 115 Professionals: physicians, nurses, n = not clear Setting: outpatient (Harvard Community Health Plan (HMO), Boston USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 20 months | |
| Outcomes | Quality of care: rate of repeat blood pressure recorded at 12 months; rate of follow‐up attempted or achieved at 12 months (median) Patient outcomes: percentage patients with blood pressure < 100 or on treatment | |
| Clinical area and targeted activity | Hypertension (general management) | |
| Reminder | Description: If there were not 2 repeat visits that included blood pressure measurement within 6 months after the initial recording of the elevated blood pressure, the COSTAR system was programmed to automatically generate a reminder notice to the patient's primary physician. The physician was notified of the deviation from the standard, and was given an encounter form on which he or she recorded when the next follow‐up visit should occur. Typology: patient‐specific: YES, space for response: YES, explicit advice: NO; explanation: NO; reference: NO; at point‐of‐care: NO | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "randomly assigned to a control or to an experimental group" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | not described |
| Incomplete outcome data (attrition bias) | Low risk | 8 dropouts in intervention group and 10 in control group |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Low risk | groups stratified by age and diastolic blood pressure level before randomization |
| Adequate protection against contamination? | High risk | physicians treated both intervention and control patients |
| Methods | Randomized trials, unit of allocation: patient | |
| Participants | Patients: 40‐60 years old; women: 68%, mean age: 51, n = ˜700 randomized (395 analyzed in included study groups) Professionals: physicians, n = not clear Setting: outpatient (University of Virginia internal medicine clinic, Charlottesville, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: nine months | |
| Outcomes | Quality of care: overall compliance rate with preventive care guidelines (primary); compliance with: dental exam, ocular pressure check, fecal occult blood test, influenza vaccination, pneumococcal vaccination, tetanus toxoid, mammography, pap smear | |
| Clinical area and targeted activity | Preventive care (multiple: test ordering, vaccination) | |
| Reminder | Description: A standardized telephone questionnaire and computer program were used to create an individualized schedule for preventive care needs. The reminder specified for the patient which services were necessary and when they should be obtained. Physician received the reminder as a memorandum appended to each patient's chart at the first visit after the telephone interview. If there was no scheduled visit, the reminder was mailed to the patient's primary physician. Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: NOT CLEAR ; reference: NO; at point‐of‐care: YES | |
| Notes | Additional study intervention excluded from analyses: patient reminder | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "eligible patients were randomly assigned to three study groups" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | data gathered through outpatient medical record review, but no mention of blinding |
| Incomplete outcome data (attrition bias) | Low risk | authors gave reasons for data excluded from analysis and compared a random sample of excluded patients with included patients |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Low risk | "The experimental groups were similar in all characteristics." |
| Adequate protection against contamination? | High risk | patients randomized |
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: women 25‐49 years old without pap smear in previous three years, n = 3052 Professionals: physicians, n = not clear Setting: outpatient (3 medical centers of a Kaiser Permanente HMO, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: not clear | |
| Outcomes | Quality of care: pap smear rate within 12 months if due (primary) Resource use/financial: total estimated costs; estimated costs per pap smear obtained | |
| Clinical area and targeted activity | Preventive care/cervical cancer (test ordering) | |
| Reminder | Description: A chart reminder was affixed to the outside of the patient's medical record. Computerized laboratory files were used to identify women without a pap smear in the previous 3 years. Typology: patient‐specific: YES; space for response: NOT CLEAR; explicit advice: NOT CLEAR; explanation: NOT CLEAR; reference: NOT CLEAR; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "randomly assigned" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Low risk | outcome data from computerized laboratory records |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Unclear risk | not reported |
| Adequate protection against contamination? | High risk | patients randomized |
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: women, ≥ 40 years old, mean age: not clear, n = 2368 women randomized (1527 analyzed in included study groups) Professionals: 20 physicians (2 family medicine physicians, 9 internal medicine physicians, 9 gynecologists) Setting: outpatient (2 sites of a large HMO, Detroit, USA) | |
| Interventions | Control (comparison 1): usual care Intervention (comparison 1): physician reminder Active control (comparison 2): patient reminder Intervention (comparison 2): physician reminder; patient reminder Duration of intervention: 12 months | |
| Outcomes | Quality of care: mammography rate if due (primary) | |
| Clinical area and targeted activity | Preventive care/breast cancer (test ordering) | |
| Reminder | Description: Electronic HMO administrative records and previous mammograms were employed to assess eligibility. The reminder forms were generated off‐site and placed in medical records by the research team. The physician reminder was a brightly colored single page notice placed in the medical chart 1 month before the due date. In addition, the reminder displayed information concerning previous mammograms and allowed the physician to recommend an alternative due date. The reminder was removed from the chart once documentation of completed mammography was obtained. Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | Data extracted from graphics. Author not contacted because publication date > 10 years. | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "randomly assigned" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Low risk | electronic administrative records were used |
| Incomplete outcome data (attrition bias) | Low risk | 2368 women randomized ‐ 741 with no visits = 1627 analyzed (68.7%) |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Low risk | "There were no significant differences among characteristics of the intervention groups" |
| Adequate protection against contamination? | High risk | patients randomized |
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: women, 18‐40 years old, mean age: not clear, n = 5801 women randomized (3848 analyzed in included study groups) Professionals: 20 physicians (2 family medicine physicians, 9 internal medicine physicians, 9 gynecologists) Setting: outpatient (three sites of a large HMO, Detroit, USA) | |
| Interventions | Control (comparison 1): usual care Intervention (comparison 1): physician reminder Active control (comparison 2): patient reminder Intervention (comparison 2): physician reminder; patient reminder Duration of intervention: 14 months | |
| Outcomes | Quality of care: pap smear rate if due (primary) | |
| Clinical area and targeted activity | Preventive care/cervical cancer (test ordering) | |
| Reminder | Description: The reminder forms were computer‐generated off‐site and placed in medical records by the research team. The physician reminder was a brightly colored single page notice that included patient specific pap smear information. The notice was prominently placed at the front of the patient's medical chart 2 months before the due date. The reminder was removed from chart once documentation of completed pap smear was obtained. Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | 2 of the 3 sites had previously participated in a related trial on reminders. | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "women were first randomly assigned, using a site specific, stratified randomization procedure to physician reminder intervention; women were then randomized to the patient reminder intervention on a weekly basis in groups of 156" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | paper copies of pap smear results received: no indication of blinding. |
| Incomplete outcome data (attrition bias) | High risk | 5801 women randomized but 3848 analyzed (66.3%) |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Low risk | characteristics of women were similar and there were no significant differences |
| Adequate protection against contamination? | High risk | patients randomized |
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: women, ≥ 40 years old, mean age: 62, n = 1262 Professionals: 30 physicians (12 faculty and 18 residents) (2 providers involved in study excluded from analyses) Setting: outpatient (Family practice center of the department of Family Medicine, University hospital, Philadelphia, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 6 months | |
| Outcomes | Quality of care: mammography rate if due (primary); mammography rate | |
| Clinical area and targeted activity | Preventive care/breast cancer (test ordering) | |
| Reminder | Description: The date of the last mammogram ordered and entered into the database was displayed in the comments section of the encounter form for each visit. The information was printed as "last mammogram: date" or, if no mammogram was on record in the database (none since 1984), the notation listed "last mammogram?" The physician ordered the mammogram by writing for one in the test ordered section of the encounter form. Typology: patient‐specific: YES; space for response: YES; explicit advice: NO; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "assigned according to a computer‐generated random number program" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | "data entered by office receptionists" |
| Incomplete outcome data (attrition bias) | Unclear risk | patients (n = not clear) from the 2 physician investigators excluded from analysis |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Low risk | some proportion of women up‐to‐date at baseline |
| Characteristics at baseline similar? | Low risk | "the subjects in the experimental and control groups were similar in age, race, insurance coverage and complexity of disease". |
| Adequate protection against contamination? | High risk | physicians likely saw patients with and without reminders |
| Methods | Cluster randomized trial, unit of allocation: health professional (stratified by level of training) | |
| Participants | Patients: ≥ 65 years old or with any of the following clinical diagnoses: diabetes, renal failure, anemia, congestive heart failure, asthma, or chronic obstructive pulmonary disease; women: 74%; mean age: not clear, n = 489 analyzed Professionals: 30 physicians (12 faculty and 18 residents) Setting: outpatient (Family practice center, Thomas Jefferson University, Philadelphia, USA) | |
| Interventions | Control: usual care Intervention: physician reminders (always) Duration of intervention: 2 months | |
| Outcomes | Quality of care: vaccination rate if due (primary) | |
| Clinical area and targeted activity | Preventive care/influenza (vaccination) | |
| Reminder | Description: "Flu vac in 1987?__" appears on the encounter form, computer‐generated for each visit when a patient arrives at the receptionist's desk, which is then attached to the front of the patient chart; it includes space for information regarding tests and procedures ordered. Typology: patient‐specific: NO; space for response: YES; explicit advice: NO; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | Additional study intervention excluded from analyses: physicians reminders (sometimes: printed for half the eligible patients) | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "Physician randomly assigned via a computerized randomization program" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | not clear |
| Incomplete outcome data (attrition bias) | Low risk | 864 patients randomized ‐ 51 received vaccination before start ‐ 93 saw multiple physicians ‐ 24 drop‐in visits = 686 analyzed (79.4%) |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Unclear risk | not reported |
| Adequate protection against contamination? | Low risk | physician assignment overruled patient assignment: physicians in control group never received a reminder |
| Methods | Cluster randomized trial (2 x 2 factorial design), unit of allocation: half‐day session, n = 30 | |
| Participants | Patients: ≥ 75 years old or ≥ 50 with serious underlying disease, women: 66%, mean age: 65, n = 1394 eligible patients (1009 analyzed), n = 1160 patient/physician pairs Professionals: 147 providers (39 faculty, 108 residents) Setting: outpatient (4 practices with 8 half‐day sessions each, Indiana University, USA) | |
| Interventions | Control: usual care Intervention (3 intervention arms combined): physician reminder Duration of intervention: not clear | |
| Outcomes | Quality of care: rate of patients who completed either directives; patient/physician pair who discussed directives | |
| Clinical area and targeted activity | Advance directives for patients at risk for acute deterioration (professional‐patient communication) | |
| Reminder | Description: All physicians routinely received computer‐generated reminders for patients with scheduled visits. They were reminded to give preventive care, note abnormal results, and avoid drug interactions. These reminders appeared at the bottom of computer‐generated printed encounter forms. The advance directive reminders were followed by a choice list (discussed today, next visit, not applicable, patient too ill, patient refuses to discuss, I disagree with advance directives). Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | Intervention a: physician reminder for instruction directive Intervention b: physician reminder for proxy directive Intervention c: physician reminder for both | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "we randomly assigned all of the physicians who worked in a particular half‐day session to the same reminder category." |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Low risk | research assistants who collected data were blinded at all times to the patient's study groups |
| Incomplete outcome data (attrition bias) | Low risk | 1394 eligible patients ‐ 42 missed by research assistant ‐ 83 low scores ‐ 36 in nursing homes or prison ‐ 29 deaf or spoke no English ‐ 9 completed advanced directive ‐ 5 for other reasons = 1190 patients interviewed and 1042 enrolled ‐ 33 patients cared for by 10 physicians who changed sessions = 1009 analyzed |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | High risk | similar, except for ethnicity and coronary heart disease and significant difference among groups for "had ever thought about advance directive": higher for control group |
| Adequate protection against contamination? | Low risk | "16 physicians practiced in more than one session per week: therefore we randomly assigned the sessions in a stepped manner by first allocating the 16 physicians and all of their associated sessions; we then randomly assigned the remaining eight sessions" |
| Methods | Cluster‐randomized trial, unit of allocation: primary clinics (n = 112) | |
| Participants | Patients: women: 37.5% , mean age: 65.6, n = 7448 Professionals: physicians, nurses, n = 600 Setting: outpatient (112 primary care clinics of a HMO, Israel) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 6‐36 months (mean: 21) | |
| Outcomes | Quality of care: rate of appropriate lipoprotein monitoring; rate of initiation or up‐titration of statin therapy; rate of up‐titration in eligible patients Patient outcomes: LDL levels, event‐free survival (only intervention group data) | |
| Clinical area and targeted activity | Cardiovascular disease (multiple: general management, prescription) | |
| Reminder | A written reminder with patient‐tailored recommendations was mailed to the primary care physicians and nurses. The recommendations were based on the previous 6 months data for new patients, and 4 months for patients in periodic follow‐up. The reminder indicated the patient’s risk factors, lipoprotein values, and known dispensed medications. Lipid‐lowering drug treatment was recommended only in patients with LDL > 110 mg/dL and consisted of either statin initiation (simvastatin 20 mg/day), statin up‐titration (doubling the last registered dose), changing to a more potent statin or compliance evaluation. For unresponsive and compliant patients it was recommended that they be referred to a metabolic clinic. Typology: patient‐specific: YES; space for response: NOT CLEAR; explicit advice: YES; explanation: YES; reference: NOT CLEAR; at point‐of‐care: NO | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | Method of randomization not described in paper. |
| Allocation concealment (selection bias) | Unclear risk | insufficient information |
| Blinding of outcome assessment (detection bias) | Low risk | administrative data. "data were collected from three routinely used databases" |
| Incomplete outcome data (attrition bias) | Low risk | The number of attrition and exclusions pre‐randomization were reported, as well as reasons for attrition/exclusions. No exclusions after randomization. |
| Selective reporting (reporting bias) | High risk | Outcomes defined in the methods are not reported in the results (initiation of statin, LDL levels in middle and low groups, and secondary outcomes (mortality and hospitalizations)) |
| Other bias | Unclear risk | not clear. |
| Outcomes at baseline similar? | Unclear risk | No baseline measures of outcomes |
| Characteristics at baseline similar? | High risk | "Patients in the intervention and control groups were similar in most parameters. However, there were significantly more patients with a history of myocardial infarction (33.0% vs. 29.9%, P = 0.004) and percutaneous coronary intervention (26.2% vs. 23.8%, P = 0.019) in the intervention arm." |
| Adequate protection against contamination? | Low risk | Clinic randomized to avoid contamination. |
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: ventricular ejection fraction < 40%; women: 0.7%; mean age: 67.5, n = 600 [277 analyzed] Professionals: not clear Setting: inpatient and outpatient (VA, Palo Alto, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 18 months | |
| Outcomes | Quality of care: ACE inhibitor use rate (primary) Patient outcomes: mean systolic/diastolic blood pressure; mean creatinine level; mortality | |
| Clinical area and targeted activity | Ventricular dysfunction (prescribing) | |
| Reminder | Description: Echocardiography reports included this statement: "Note: patients with ejection fraction < 40% have a survival benefit with ACE inhibitors (goal dose lisinopril or fosinopril 30‐40 mg/day)". Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: YES; reference: NO; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "Patient selection and randomization were computerized and performed in conjunction with the generation of the reminder: patients who met study entry criteria were randomized using a computerized random number generator." |
| Allocation concealment (selection bias) | Low risk | "Allocation was concealed from all echocardiographers until the reminder appeared on the report." |
| Blinding of outcome assessment (detection bias) | Unclear risk | "medication use was determined from review of inpatient or outpatient encounters" |
| Incomplete outcome data (attrition bias) | Low risk | Among the 600 eligible and randomized patients, 201 were excluded from analyses as they were already on ACE‐inhibitor or appropriate alternative at the time of randomization (96 [32.9%] in intervention group and 105 [34.1%] in control group), 46 died within 2 months of the echocardiogram (20 [6.8%] in intervention group and 26 [8.4%] in control group), 71 left the VA healthcare system (36 [12.3%] in intervention group and 35 [11.4%] in control group) , and 5 had an allergy or adverse reaction to ACE‐inhibitors (3 [1%] in intervention group and 2 [0.6%] in control group). In total, 323 [53.8%] were excluded from analyses (155 [53.1%] in intervention group and 168 [54.5%] in control group). |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Low risk | no significant difference between groups for ACE inhibitor use |
| Characteristics at baseline similar? | Low risk | no significant difference except for history of heart failure, but not extreme. Adjusted for in multivariate analyses. |
| Adequate protection against contamination? | High risk | patients randomized rather than physicians |
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: men: 98%, mean age: 69, n = 1546 [1271 analyzed] Professionals: physicians (n = 45) and nurse practitioners (n = 5), n = 50 Setting: outpatient and inpatient (VA, Palo Alto, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 56 months | |
| Outcomes | Quality of care: prescription rate of any beta‐blocker within 9 months (primary); prescription rate of recommended beta‐blockers within 9 months | |
| Clinical area and targeted activity | Ventricular dysfunction (prescribing) | |
| Reminder | Description: In the process of printing the completed echocardiography report, the computer algorithm checked the electronic report and had a reminder attached to the report of eligible patients meeting the pre‐defined criteria. The reminder included the following statement: "Note: Patients with reduced left ventricular ejection fraction have a survival benefit with beta‐blockers (initial dose: carvedilol 3.125 mg BID or metoprolol succinate 12.5 mg BID)". The reminder also recommended cardiology follow‐up if the patient had New York Heart Association class II or IV symptoms. Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: YES; reference: NO; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "Randomization was carried out in conjunction with the formation of the reminder with an electronic database: patients meeting study criteria were randomized with a computerized random number generator. Randomization was performed separately at each site" |
| Allocation concealment (selection bias) | Low risk | "Allocation was concealed from all echocardiographers until the reminder appeared on the report." |
| Blinding of outcome assessment (detection bias) | Low risk | "prescription was determined with the VA pharmacy database" |
| Incomplete outcome data (attrition bias) | Low risk | See Figure 1: Among the 1546 eligible and randomized patients, 89 died within 30 days of the echocardiogram (44 [5.8%] in intervention group and 45 [5.7%] in control group), 180 left the healthcare system (88 [11.7%] in intervention group and 92 [11.6%] in control group) , and 6 had an echocardiography at >1 site (2 [0.3%] in intervention group and 4 [0.5%] in control group). In total, 275 [17.8%] were excluded from analyses (134 [17.7%] in intervention group and 141 [17.8%] in control group). Exclusions did not significantly differ between groups. |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Low risk | no significant difference between groups for β‐Blocker use |
| Characteristics at baseline similar? | Low risk | no significant difference except for percentage of male, but not extreme. |
| Adequate protection against contamination? | High risk | patients randomized rather than physicians. |
| Methods | Cluster‐randomized trial, unit of allocation: clinic, n = 5 | |
| Participants | Patients: ≥ 70 yr or ≥ 50 yr with severe chronic illness, without advanced directives; women: 68%; mean age: 72; n = 719 in included study arms Professionals: 31 providers in included study arms Setting: outpatient (5 clinics in 5 general practices in Boston, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 7 months | |
| Outcomes | Quality of care: completion rate of any advanced directives (primary); discussion or completion rate of advanced directives; completion rate of healthcare proxy; completion rate of living will | |
| Clinical area and targeted activity | Advance directives (professional‐patient communication) | |
| Reminder | Description: Reminders were printed at the bottom of patient summary sheets at every patient visit during the study period. It read: "Your patient is x years old with a history of y medical condition and is missing key information about advanced directives". It instructed physicians to enter data about discussion or completion of the living will and healthcare proxy in either the inpatient or the outpatient electronic medical record. Reminders were already in use for screening tests. Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | Additional study intervention excluded from analyses: physician reminder; patient reminder; patient educational material | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "A blinded programmer using a random number generator randomly assigned each group" |
| Allocation concealment (selection bias) | Low risk | "A blinded programmer using a random number generator randomly assigned each group" |
| Blinding of outcome assessment (detection bias) | High risk | "The study investigator who assessed outcomes was not blinded". |
| Incomplete outcome data (attrition bias) | Low risk | Among the 950 randomized patients, 41 were ineligible (17 [3.4%] in intervention group and 24 [5.4%] in control group) and 190 refused to participate (102 [20.2%] in intervention group and 88 [19.7%] in control group). In total, 719 [75.7%] were included in the intention‐to‐treat analyses (385 [76.4%] in intervention group and 334 [74.9%] in control group). |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | Not reported |
| Characteristics at baseline similar? | High risk | significant differences at baseline for age, gender, private insurance, and number of qualifying illnesses |
| Adequate protection against contamination? | Low risk | physicians randomized |
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: all members of a managed care plan, 12‐64 years old; women: 58%; mean age: 38, n = 41,870 [35,447 completed study] (n = not clear for included analyses) Professionals: not clear Setting: mixed (managed care plan, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 12 months | |
| Outcomes | Quality of care: compliance rate with starting a new drug (primary) (other outcomes not available) Resource use/cost: mean hospital admissions; mean inpatient days; mean length of stay; inpatient charges | |
| Clinical area and targeted activity | General management: increase compliance with evidence‐based practices (prescribing) | |
| Reminder | Description: The sentinel system is designed as a rule‐based artificial intelligence engine combined with an automated message generator that conveys clinical recommendations and supporting literature to treating physicians. Daily data inputs include physician‐generated insurance claims, hospital discharge and outpatient claims, laboratory claims and laboratory test results, and pharmacy claims. Typical issues targeted by the rules engine include the following: a) absence of ACE inhibitor therapy in patients with congestive heart failure and in those who meet the HOPE trial criteria; b) absence of β‐blocker use in patients with myocardial infarction; c) absence of anticoagulation in patients with atrial fibrillation and structural heart disease; d) absence of documented laboratory monitoring in patients taking warfarin sodium, glitazones, and other medications that require specific laboratory tests. The system contains more than 1000 decision matrices that, when triggered, result in the transmission of a communication to the treating physician (on paper). All recommendations make clear that the communication is merely for the physician's consideration and that there may be mitigating circumstances that might render the recommendation inappropriate. Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: NOT CLEAR; reference: NO; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "Patients were assigned to an intervention or a control group, using an individually assigned random number. Assignment occurred on a single date at study entry." |
| Allocation concealment (selection bias) | Low risk | "Assignment occurred on a single date at study entry. Neither patients nor treating physicians were informed of the allocation, although it is likely that physicians who received communications about specific patients surmised that those patients were part of the intervention group." |
| Blinding of outcome assessment (detection bias) | Unclear risk | Not specified in article. Likely from a computer database. |
| Incomplete outcome data (attrition bias) | Low risk | 41870 patients randomized ‐ 2408 withdrew before study inception (1206 [5.9%] in control group and 1202 [5.8%] in intervention group) ‐ 4015 withdrew between 6 and 12 months (2088 [10%] in control group and 1927 [9.2%] in intervention group) = 35447 [84.7%] completed study (17635 [84.3%] in control group and 17812 [85.1%] in intervention group) |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Low risk | "at baseline, no significant differences were observed in age or sex between the intervention and control group subjects" |
| Adequate protection against contamination? | High risk | physicians likely care for patients in intervention and control group |
| Methods | Cluster‐randomized trial, unit of allocation: health professional (n = 144) | |
| Participants | Patients: age 50‐74, registered with a study provider and visited a study provider during study period; women: 55.8%; median age: 60; n = 20788 Professionals: 144 general practitioners Setting: outpatient (individual or group primary care practices, Val‐de‐Marne, France) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 17 months | |
| Outcomes | Quality of care: unadjusted patient adherence to colorectal cancer screening (completion of FOBT or exclusion from FOBT for medical reason), adjusted patient adherence (adjusted for clustering) | |
| Clinical area and targeted activity | Preventive care/colorectal cancer (test ordering) | |
| Reminder | The screening centre mailed three reminders to the intervention‐group GPs at 4‐month intervals. The reminders were lists of patients who had not performed a scheduled FOBT. Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: NO; reference: NO; at point‐of‐care: NO | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "GPs were assigned randomly in a 1:1 proportion to the intervention or the control group, in permuted blocks of 2 or 4" |
| Allocation concealment (selection bias) | Low risk | unit of allocation by professional (GP) |
| Blinding of outcome assessment (detection bias) | Low risk | "Outcome data were collected for all patients in the ADOC94 database throughout the study period and over the 9‐month period following the last reminder by data abstracters who were blinded to group assignment." |
| Incomplete outcome data (attrition bias) | Low risk | "We included 20 778 patients who were eligible for CRC screening and had visited a study GP for any reason during the study period". No missing data. See Figure 1 in paper. |
| Selective reporting (reporting bias) | Low risk | Primary outcome in trial register matches the primary outcome in the published paper. |
| Other bias | Unclear risk | Low GP participation rate. Only 15% of contacted GPs agreed to participate in the study. |
| Outcomes at baseline similar? | Low risk | No SS difference at baseline. "CRC screening adherence rates were 21.8% (95% CI 18.7–26.1) in the intervention group and 21.9% (95% CI 18.9–26.5) in the control group." |
| Characteristics at baseline similar? | High risk | Some SS differences. "Baseline GP characteristics were similar in the two groups, except for numbers of GPs in group practice and charging above‐standard fees, which were higher in the control group. The only significant difference in baseline patient characteristics was a slightly higher proportion of patients from deprived areas in the intervention group." |
| Adequate protection against contamination? | Low risk | Randomization at the GP level. Low risk of contamination. |
| Methods | Cluster‐randomized trial, unit of allocation: health professional | |
| Participants | Patients: diabetic; n = 497 eligible patients (359 analyzed); encounters: 1265 (884 analyzed) Professionals: 58 primary care providers randomized (20 family physicians, 1 general internist, 2 physician's assistants, 2 nurse practitioners, 33 residents) [30 analyzed] Setting: outpatient (Duke Family Medicine Center, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: six months | |
| Outcomes | Quality of care: clinician compliance rate overall (number of recommendations completed/total number of recommendations) (primary); clinician compliance rate for 8 recommendations: foot exam every month in patients with diabetic neuropathy or history of lower limb ulcer, annual complete physical exam, glycated hemoglobin every 6 months, annual urine protein determination, annual cholesterol level, annual eye exam, seasonal influenza vaccination, pneumococcal vaccination | |
| Clinical area and targeted activity | Diabetes (multiple: test ordering, vaccination) | |
| Reminder | Description: The Computer‐Assisted Management Protocol generates a set of disease‐specific care recommendations customized to an individual patient, based on data stored in the patient's electronic medical record, that advises the clinician regarding which studies/procedures should be done during the current visit and which studies/procedures are next due in order to assist the clinician with managing the diabetic patient in accordance with a clinical practice guideline. The output is printed on the 1st page of the paper encounter form. Additional flexibility is included for the clinicians to designate that the recommendation was declined by the patient ("D") or never to be done for the patient ("N"). Clinicians can order the appropriate studies/procedures on the encounter form and indicate results of procedures done during the encounter. Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: YES; reference: NO; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "all primary care providers were randomly assigned by standard randomization techniques" |
| Allocation concealment (selection bias) | Low risk | "all primary care providers were randomly assigned by standard randomization techniques" |
| Blinding of outcome assessment (detection bias) | Low risk | "chart audit with audit protocol: intra‐auditor consistency > 90%" |
| Incomplete outcome data (attrition bias) | Low risk | |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Low risk | "compliance scores prior to study were not statistically significantly different" |
| Characteristics at baseline similar? | Low risk | "patients did not differ significantly by age, race or gender" |
| Adequate protection against contamination? | Low risk | providers randomized |
| Methods | Cluster‐randomized trial; unit of allocation: health professional | |
| Participants | Patients: patients with heart failure or with ischemic heart disease; women: 60%; median age: 75; n = 171 Professionals: primary care physicians, n = 769 potentially eligible randomized (128 analyzed) Setting: outpatient (Alberta, Canada) | |
| Interventions | Active control: medication profile Intervention: physician reminder; medication profile Duration of intervention: 3.5 year | |
| Outcomes | Quality of care: prescription rate of efficacious therapies within 6 months (primary); prescription rate of ACE inhibitors or ARBs in heart failure patients; prescription rate of statins in ischemic heart disease patients | |
| Clinical area and targeted activity | Cardiovascular disease (prescribing) | |
| Reminder | Description: For each patient, a condition‐specific one‐page evidence summary was generated in the form of a letter addressed to the primary care physician and was faxed to him. The letters identified patients and their diagnoses, briefly described the key evidence in support of the study medications, and were signed by opinion leaders. The intent of the intervention was that the evidence summary (and medical profile) would become part of the patient's medical record and act as a point‐of‐care reminder for the next patient visit. Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: YES; reference: YES; at point‐of‐care: NO | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "simple randomization with concealment of allocation was performed with the use of a computer‐generated sequence" |
| Allocation concealment (selection bias) | Low risk | "simple randomization with concealment of allocation was performed with the use of a computer‐generated sequence" |
| Blinding of outcome assessment (detection bias) | Low risk | "all outcomes were ascertained in an independent and blinded fashion, and allocation was concealed from patients, investigators, data collectors, and analysts" |
| Incomplete outcome data (attrition bias) | Low risk | "no study patients were lost to follow‐up". 769 professionals were randomized, but 128 were analyzed as most did not contribute an eligible patient. |
| Selective reporting (reporting bias) | High risk | results on secondary outcomes not reported. |
| Other bias | Low risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Low risk | "the intervention and control subjects were comparable, with no important differences" |
| Adequate protection against contamination? | Low risk | physicians allocated to intervention |
| Methods | Cluster non‐randomized trial; unit of allocation: clinical area, n = 4 | |
| Participants | Patients: patients with non‐insulin dependent diabetes mellitus; women: not clear; mean age: not clear; total n = 2791 (estimated 1395 in the 2 included study arms) Professionals: 99 internal medicine residents and 15 faculty internists (total n = 114) in four clinical areas (estimated 57 in the 2 included study arms) Setting: outpatient (4 clinic areas of a general medicine clinic, Indiana University, USA) | |
| Interventions | Active control: educational meeting (postgraduate seminar) Intervention: physician reminder; educational meeting (postgraduate seminar) Duration of intervention: 11 months | |
| Outcomes | Quality of care: adherence rate to recommendations for: lab orders (glycated hemoglobin, fasting blood glucose) and therapies (home‐monitored blood glucose, diet clinic referral, oral hypoglycemic agents) (median) | |
| Clinical area and targeted activity | Diabetes (multiple: test ordering, prescribing, referral) | |
| Reminder | Description: Printed reminders were placed in patients' clinic records whenever the computer detected history, physical, laboratory, or pharmacy data indicating the need to consider a recommendation (e.g. if the patient was obese and without a diet on record, the computer would print out the following reminder on a separate sheet). (same system as McDonald 1984) Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: NOT CLEAR; reference: NO; at point‐of‐care: YES | |
| Notes | Additional study interventions excluded from analyses: a) physician reminder; educational meeting (postgraduate seminar); physician educational material; b) physician reminder; educational meeting (postgraduate seminar); physician educational material; patient education service | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | High risk | "the four areas were assigned arbitrarily to study conditions according to a nonequivalent control group design" |
| Allocation concealment (selection bias) | High risk | "the four areas were assigned arbitrarily to study conditions according to a nonequivalent control group design" |
| Blinding of outcome assessment (detection bias) | Low risk | "computerized audit of medical record" |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Low risk | "baseline interviews with a random sample of 175 patients showed the 4 groups to be equivalent with respect to general therapeutic practices" |
| Characteristics at baseline similar? | Unclear risk | not clear |
| Adequate protection against contamination? | Low risk | randomization by clinical area |
| Methods | Cluster‐randomized trial, unit of allocation: health professional | |
| Participants | Patients: > 18 years old, diagnosed with coronary artery disease (stenosis in at least one coronary vessel of ≥ 50%) eligible for but not already taking a statin or who were on a suboptimal regimen, women: 21%, mean age: 64, n = 480 Professionals: primary care physicians, n = 252 Setting: outpatient (252 general practices in Alberta, Canada) | |
| Interventions | Active control: coronary artery diagram Intervention (two intervention arms combined): coronary artery diagram; physician reminder Duration of intervention: 6 months | |
| Outcomes | Quality of care: compliance with recommendation at six months (statin initiation or increase in dose) (primary); taking a statin; standardized statin dose; taking another lipid‐lowering drug; acetylsalicylic acid; acetylsalicylic acid or thienopyridine; ACE inhibitor; ACE inhibitor or ARB; β‐Blocker; triple therapy Patient outcomes: mortality rate Resource use/cost: ED visits, hospitalizations | |
| Clinical area and targeted activity | Cardiovascular disease (prescribing) | |
| Reminder | Description: The statement was a 1‐page summary of evidence‐based secondary prevention strategies and treatment recommendations. The Local Opinion Leader Statement contained the signatures of 5 local opinion leaders, while the Unsigned Evidence Statement was unsigned. These statements were imprinted with the name of the patient, addressed directly to the patient's physician, and faxed automatically by a software program, along with a coronary artery diagram documenting the extent of the patient's coronary atherosclerosis. These statements were sent to physicians within a few days of the angiogram. Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: YES; reference: YES for intervention a/NO for intervention b; at point‐of‐care: NO | |
| Notes | Intervention a: coronary artery diagram; physician reminder (local opinion leader statement) Intervention b: coronary artery diagram; physician reminder (unsigned evidence statement) Intervention groups considered separately in 1 analysis (content of reminder: reference versus no reference) | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "computer‐generated central randomization system with concealment of the randomization list" |
| Allocation concealment (selection bias) | Low risk | "computer‐generated central randomization system with concealment of the randomization list" |
| Blinding of outcome assessment (detection bias) | Low risk | "Blinding of outcome assessors and analysts" |
| Incomplete outcome data (attrition bias) | Low risk | reasons provided for excluded patients; intention‐to‐treat analyses |
| Selective reporting (reporting bias) | Low risk | outcomes reported match outcomes described in protocol |
| Other bias | Low risk | no recruitment bias; GEE analyses |
| Outcomes at baseline similar? | Low risk | "No appreciable differences between the treatment arms" for statin use (P = 0‐87) and statin dosing (P = 0.84) |
| Characteristics at baseline similar? | Low risk | "At baseline, there were no statistically significant differences between groups". |
| Adequate protection against contamination? | Low risk | randomization by health professional |
| Methods | Cluster non‐randomized trial (cross‐over), unit of allocation: health professional | |
| Participants | Patients: percentage of women: not clear, mean age: not clear, n = 189 Professionals: physicians, n = 9 Setting: outpatient care (general medicine clinic at Wishard Memorial Hospital, Indianapolis, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 17 weeks | |
| Outcomes | Quality of care: overall compliance rate (primary); compliance rate with observing a physical finding or inquiring about a symptom; compliance with ordering a diagnostic study; compliance with changing or initiating a therapeutic regimen | |
| Clinical area and targeted activity | General care (multiple: prescribing, test ordering, other) | |
| Reminder | Description: The Regenstrief Medical records system searches its records for events and makes recommendations about the management. Each recommendation consists of a reminder to the physician that a particular event has occurred and a suggested course of "action" for correcting that event. The study involved 390 protocols (recommendations). The computer prints 3 reports. The first is the surveillance report, which contains all the computer recommendations for a given patient. The second is a computer‐tailored encounter form, which provides space for recording findings. The third report, the summary report, is not influenced by the computer protocols and is a flow‐sheet summary. Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: YES; reference: NO; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | High risk | no mention of randomization |
| Allocation concealment (selection bias) | High risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | outcome data gathered through medical record review, no indication of blinding |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Unclear risk | not reported |
| Adequate protection against contamination? | High risk | cross‐over design |
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: adult diabetic patients; percentage of women: not clear, mean age: 60, n = 257 (226 analyzed) Professionals: diabetologists, residents, interns, senior medical residents, nurse clinicians, n = 63 Setting: outpatient (diabetic clinic of Wishard Memorial Hospital, Indianapolis, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 8 months | |
| Outcomes | Quality of care: compliance rate with recommendations for test ordering/therapeutical change (median) | |
| Clinical area and targeted activity | Diabetes (multiple: prescribing, test ordering) | |
| Reminder | Description: The computerized medical record system (Regenstrief) generated prospective, protocol‐driven recommendations, which alerted the clinician to the existence of, and the proper response to, simple events. Two types of protocols: 1) taking measurement at proper intervals, based on "if on drug A and no test B for X months then order test B"; 2) reacting to measures that implied that treatment was either insufficient, excessive, or dangerous, based on "if on drug A and last test B abnormal, then warn about possible changed drug action". The computer printed 3 reports: the summary report, with a flow‐sheet; the patient encounter form with space for writing new medication orders; the surveillance report, with the protocol‐generated suggestions to physician, with recommended tests, date of last test, treatments triggering recommendation, recommendations for specific changes in therapeutics, with rationale for the change suggested. Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: YES; reference: YES; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | randomized by computer‐generated random number |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | not clear |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Unclear risk | not reported |
| Adequate protection against contamination? | High risk | most practitioners saw patients from both the study and control groups |
| Methods | Cluster‐randomized trial (cross‐over trial), unit of allocation: health professional | |
| Participants | Patients: n = not clear; conditions: 3691 Professionals: 31 providers (9 interns, 17 residents, 5 nurse‐practitioners) Setting: outpatient (General medicine service, University Hospital, Indianapolis, USA) | |
| Interventions | Control: usual care Intervention (2 intervention arms combined): physician reminder Duration of intervention: 15 weeks | |
| Outcomes | Quality of care: compliance rate | |
| Clinical area and targeted activity | Preventive care (multiple: test ordering, prescribing, general management) | |
| Reminder | Description: At each visit, the computer provides a tailored encounter form and a flow sheet summary of each patient's medical history and follows physician‐authored management rules (n = 410) to remind the physician about patient conditions requiring his attention. The report containing these reminders is called the "surveillance" report. Practitioners order all diagnostic treatments and referrals by recording them on the encounter form. Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: YES; reference: YES for intervention a / NO for intervention b; at point‐of‐care: YES | |
| Notes | Intervention a: physician reminder (with bibliographic citations) Intervention b: physician reminder (without citations) Intervention groups considered separately in 1 analysis (content of reminder: reference versus no reference) | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | physicians were randomly assigned to the order to which they received their intervention |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | not clear |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Unclear risk | not reported |
| Adequate protection against contamination? | High risk | cross‐over trial |
| Methods | Cluster‐randomized trial, unit of allocation: practice team, n = 27 | |
| Participants | Patients: women: 65%; mean age: not clear; n = 12,467 Professionals: 130 providers (115 residents, 11 faculty, 4 nurses) within practice teams (n = 115 in included study groups) Setting: outpatient (27 teams in a general medicine service, University Hospital, Indianapolis USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 2 years | |
| Outcomes | Quality of care: overall compliance rate (primary); compliance with occult blood, cervical smear, chest roentgenogram, pneumococcal vaccine, tuberculosis skin test, serum potassium, mammogram, influenza vaccine, diet, digitalis, antacids, beta blockers (12 reminders) | |
| Clinical area and targeted activity | Preventive care (multiple: vaccination, test ordering, prescribing) | |
| Reminder | Description: The computerized medical record was programmed with 1491 rules that could generate 751 different reminder messages. The set of rules included reminders about preventive care, obtaining tests needed to complete the initial database or to identify the cause of existing abnormalities, about prophylactic treatment and treatment of active problems. The computer used these rules to review each patient's electronic medical record the day before each visit. For each patient, when it found conditions satisfying the reminder rule, a reminder was stored on a disc file. The computer gathered these messages as a printed report that was attached to the charts of scheduled patients. The reminder messages included citations to the relevant medical literature. Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: YES; reference: YES; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "Each team was randomized as to study or control" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Low risk | outcomes measured using computer |
| Incomplete outcome data (attrition bias) | Low risk | physicians with fewer than 100 reminder messages during study period excluded (2%) |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | Not reported |
| Characteristics at baseline similar? | Low risk | No significant difference in gender and age for patients |
| Adequate protection against contamination? | Unclear risk | practice team randomized, but faculty members and nurse‐clinicians saw both study groups |
| Methods | Cluster‐randomized trial, unit of allocation: health professional | |
| Participants | Patients: ≥ 40 years old, women: 67%, mean age: not clear; n = ˜1291 patients in four included study arms Professionals: 42 residents in four included study arms Setting: outpatient (General Internal Medicine Practice, University hospital, San Francisco USA) | |
| Interventions | Control (comparison 1): usual care Intervention (comparison 1): physician reminder Active control (comparison 2): distribution of educational material to patients; patient reminder Intervention (comparison 2): physician reminder; distribution of educational material to patients; patient reminder Duration of intervention: nine months | |
| Outcomes | Quality of care: physician compliance rate with: FOBT, rectal exam, sigmoidoscopy, pap smear, pelvic exam, breast exam, and mammography (median) Cost: cost per patient; cost per additional screening test | |
| Clinical area and targeted activity | Preventive care/colorectal, breast, cervical cancer (test ordering) | |
| Reminder | Description: The cancer screening reminders provided residents with up‐to‐date information about their patient's screening status. The research staff printed cancer screening reminders at the time of each patient appointment and updated existing microcomputer files by re‐auditing patient medical records in preparation for return appointments. At the time of a patient's visit to the resident, a cancer screening reminder was attached to the regular encounter form or medical record. A new reminder was generated for each patient encounter. Reminders displayed the list of appropriate cancer screening procedures (based on the patient's sex and age), the recommended testing intervals, the last performance date, the due date for each next test, and the patient's current "due" status. The physician indicated on the form whether or not each test was performed or ordered during the current visit. Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: NO; reference: YES; at point‐of‐care: YES | |
| Notes | Data extracted from graphics. Author not contacted as publication date > 10 years. Additional study interventions excluded from analyses: a) audit and feedback, b) audit and feedback; distribution of educational material to patients; patient reminder | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "residents were randomly assigned to one of six intervention groups" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Low risk | Medical records audit: no blinding but reliability test performed on random sample (90‐98%) |
| Incomplete outcome data (attrition bias) | High risk | 1969 records audited at baseline = 72% of eligible patients; 1936 records audited during study = 71% of eligible patients |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Low risk | compliance scores at baseline not significantly different |
| Characteristics at baseline similar? | Unclear risk | not reported |
| Adequate protection against contamination? | Low risk | providers randomized |
| Methods | Non‐randomized trial, unit of allocation: patient | |
| Participants | Patients: pregnant women; mean age: not clear, n = 279 Professionals: physicians, n = 5 Setting: outpatient (Harvard Community Health Plan ambulatory care center, Boston, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 18 months | |
| Outcomes | Quality of care: compliance rate with 5 minimum standard of care (at 18months): blood group and type, syphilis serology, prenatal counseling, pregnancy diet counseling, sickle cell preparation (median) | |
| Clinical area and targeted activity | Prenatal care (multiple: test ordering, professional‐patient communication) | |
| Reminder | Description: COSTAR was programmed to automatically check the records of patients at the time of each prenatal visit to determine if physicians had complied with specific indices of care. An updated summary record was printed prior to the patient's visit with the list of missing items, under the heading "Data incomplete". Typology: patient‐specific: YES; space for response: NO; explicit advice: NO; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | High risk | "patients with odd‐numbered medical records were assigned to the experimental group (...) patients with even‐numbered medical records were assigned to a control group". |
| Allocation concealment (selection bias) | High risk | allocation based on odd/even number of medical record |
| Blinding of outcome assessment (detection bias) | Unclear risk | outcomes likely assessed by computer, but no specific indication in text |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Unclear risk | not reported |
| Adequate protection against contamination? | High risk | physicians saw patients from intervention and control groups |
| Methods | Cluster‐randomized trial (blocked by site and level of training), unit of allocation: health professional | |
| Participants | Patients: women: not clear; mean age: not clear; n = 480 identified (164 analyzed) Professionals: internal medicine residents, n = 35 randomized out of 36 Setting: outpatient (2 clinics, USA) | |
| Interventions | Active control: blank encounter form; educational meeting Intervention: physician reminder; blank encounter form; ; educational meeting Duration of intervention: 6 months | |
| Outcomes | Quality of care: overall compliance rate (primary) | |
| Clinical area and targeted activity | Diabetes (multiple: test ordering, prescribing, general management, professional‐patient communication) | |
| Reminder | Description: The computer program outputs a printed paper health maintenance report for the patient's primary physician based on the currently available data for the patient. The report summarizes the patient's diabetes preventive‐health status, and lists a schedule of upcoming or past due preventive‐health activities for the patient. Clinical alerts about high‐risk aspects of the patient's current profile are also presented. The report is placed at the front of the patient's chart. Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: NOT CLEAR; reference: NO; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "resident physicians were randomly assigned" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | manual chart review ‐ no blinding procedure described |
| Incomplete outcome data (attrition bias) | High risk | 480 patients identified, but 164 analyzed (34.2%). No exclusion reasons. |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | High risk | baseline compliance lower in control group (34.6%) than intervention group (38.0%) |
| Characteristics at baseline similar? | Unclear risk | not reported |
| Adequate protection against contamination? | Low risk | health professional randomized |
| Methods | Non‐randomized trial, unit of allocation: patient | |
| Participants | Patients: women: not clear; mean age: not clear, n = 120 Professionals: nurses (n = 109) Setting: inpatient (2 intensive care units in LDS Hospital, Department of Critical Care, Utah, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 59 days (45 days in control group) | |
| Outcomes | Quality of care: mean deficiencies per day per patient (patient's total number of deficiencies/number of valid days in study) (primary) | |
| Clinical area and targeted activity | Intensive care (multiple: record keeping, general management) | |
| Reminder | Description: The computer system stored any mid‐day (13:00) reminders generated in a database. For each patient in the study group, the program generated a reminder report which listed patient's room, patient number/name, the date and any reminder (e.g. no Glasgow Coma Score between 7:00 and 13:00). Any reports containing reminders were delivered to charge nurse, who delivered them to the bedside nurse. Typology: patient‐specific: YES; space for response: NO; explicit advice: NO; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | High risk | patients assigned according to ICU room to which they were admitted |
| Allocation concealment (selection bias) | High risk | patients assigned according to ICU room to which they were admitted |
| Blinding of outcome assessment (detection bias) | Low risk | outcome measure collected from computer system |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Unclear risk | not reported |
| Adequate protection against contamination? | High risk | 40 nurses cared for study and control patients |
| Methods | Cluster‐randomized trial, unit of allocation: practice, n = 4 | |
| Participants | Patients: > 18 years old, due for prevention services; women: 61%; mean age: 40; n = 7,397 (3564 in comparison 1; 3833 in comparison 2) Professionals: 49 family physicians (6 faculty, 1 fellow, 42 residents) Setting: outpatient (four practice groups in the Family Medicine Center at the Medical University of South Carolina, USA) | |
| Interventions | Active control (comparison 1): educational meeting; audit and feedback Intervention (comparison 1): physician reminder; educational meeting; audit and feedback Active control (comparison 2): patient reminder; educational meeting; audit and feedback Intervention (comparison 2): physician reminder; patient reminder; educational meeting; audit and feedback Duration of intervention: 12 months | |
| Outcomes | Quality of care: physician adherence rate to 5 preventive services: FOBT, mammography, tetanus vaccine, cholesterol, pap smear (median) | |
| Clinical area and targeted activity | Preventive care/colorectal, breast, cervical cancer, tetanus vaccination and cholesterol prevention (multiple: test ordering, vaccination) | |
| Reminder | Description: Reminder forms were generated by the computer system for each patient the night before a scheduled appointment. The reminders were generated by scanning each patient record for deficient preventive services based on the patient's age, sex, and last recorded time of the service. Forms were printed on single sheets of paper and attached to the medical record by nursing personnel the morning of the scheduled visit. The top half of the form listed identifying information and 0 to 5 deficient preventive services. It contained boxed for the physician to mark, indicating his or her action on each particular reminder. Actions included ordering the preventive service that day, scheduling the patient to return for it another day, noting that it was not indicated for the patient, offering it to the patient but having the patient refuse, or not discussing it. The bottom half of the reminder form listed any of the 5 preventive services appropriate for the patient's age and sex, and the date the item was last received. Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "patients and their physicians were randomly assigned by practice group into study groups" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Low risk | "[outcome] assessed through computerized medical records" |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | High risk | adherence rate at baseline differ according to study groups |
| Characteristics at baseline similar? | High risk | "statistically significant differences between study groups were present for race, insurance coverage, and visit frequency" |
| Adequate protection against contamination? | Low risk | physicians randomized |
| Methods | Cluster‐randomized trial, unit of allocation: family, n = 4450 families randomized | |
| Participants | Patients: 15 years old or more; percentage of women: not clear; mean age: not clear; 1403 patients (1056 families) randomized to usual care and 1471 patients (1122 families) randomized to included study arm Flu arm: 822 families randomized, 939 patients; blood pressure arm: 4247 families randomized, 5744 patients; Pap arm: 1406 women randomized; Tetanus arm: 4247 families randomized, 5589 patients Professionals: staff physicians, residents, and nurses, n = not clear Setting: outpatient (four practices, University of Ottawa Family Medicine Center, Canada) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 12 months (69 days for flu arm) | |
| Outcomes | Quality of care: rate of eligible patients for whom the recommended procedure was performed (primary); rate of eligible patients with: influenza vaccination, blood pressure reading, pap test, tetanus vaccination | |
| Clinical area and targeted activity | Preventive care/influenza vaccination, blood pressure, smoking status, pap test, tetanus (multiple: test ordering, vaccination) | |
| Reminder | Description: Computer‐generated reminders were included on the routinely printed encounter form before any visit to the office to remind the physician of outstanding preventive procedures. Until the procedure or reading was recorded, the computer continued to generate reminders on subsequent visits. Influenza vaccination reminder: for patients who had not already been vaccinated, the message stated "Patient 65 or older: check flu immunization". Blood pressure reminder: When the patient booked an appointment, the computer printed a reminder advising the doctor to "Check blood pressure". Pap reminder: for women who had not been screened during previous year, the computer printed a message to the physician to recommend cervical screening. Tetanus vaccination reminder: "ask patient about tetanus vaccination" was included on the routinely printed encounter form. Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | Additional study interventions excluded from analyses: a) patient phone reminder, b) patient letter reminder | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "with the use of a standard randomization computer program" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | Influenza vaccination recorded in the computer, but data completed by phone calls which are not described as blinded |
| Incomplete outcome data (attrition bias) | Low risk | possible that patients had undergone preventive measures somewhere else, but similar situation across groups |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Low risk | no significant difference for age and gender (family size was different before exclusion of practices) |
| Adequate protection against contamination? | Unclear risk | families (patients) randomized; physicians likely saw both study and control patients |
| Methods | Cluster‐randomized trial, unit of allocation: health professional | |
| Participants | Patients: men: 96%, mean age: 68, n = 719 patients Professionals: physicians (n = 15), residents/fellows (n = 44) and nurse practitioners (n = 12) Setting: outpatient (General internal medicine clinic of the VA Pounget Sound Health Care System, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 6 months | |
| Outcomes | Quality of care: prescription change rate (primary) Patient outcomes: mean blood pressure Resource use: mean clinic visits per patient, mean ED visits per patient, mean creatinine tests per patient, mean total cholesterol tests per patient | |
| Clinical area and targeted activity | Hypertension (prescribing) | |
| Reminder | Description: An automated computer query system identified eligible patients and their providers. For each clinic visit, for each eligible patient, providers had a 1‐page guideline reminder placed in the patient chart by the clinic pharmacist, attached to the medication refill forms that are given to providers at every patient visit. The reminder highlighted the prescription and offered alternative drugs and doses. For continued medication use, the reminder also asked provider to designate 1 of 4 indications. Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: YES; reference: YES; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "A random number generator was used to randomize providers, stratified to whether they were staff physicians, nurse practitioners, or residents or fellows". |
| Allocation concealment (selection bias) | Low risk | randomization at the physician level; providers were assigned numeric codes and study investigators were blinded to the coding identifiers |
| Blinding of outcome assessment (detection bias) | Low risk | outcome data from computer database |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | High risk | provider characteristics (type, gender, number patients, number patient visits) were not statistically different; patient characteristics (age, race, gender, weight, blood pressure) were not statistically different, but mean prescriptions per patient and percentage on ß‐blockers and ACE inhibitors were different |
| Adequate protection against contamination? | Low risk | physicians randomized |
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: diabetic patients; percentage of women: not clear, mean age: not clear; n = 185 (133 analyzed) Professionals: physicians, n = not clear Setting: outpatient (University of Texas Medical School Ambulatory Clinic, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 12 months (not finished) | |
| Outcomes | Quality of care: compliance rate (primary) Resource use/cost: percentage of patients hospitalized, mean days of hospitalization, mean number of patient visits, mean costs per patient | |
| Clinical area and targeted activity | Diabetes (multiple: not specified) | |
| Reminder | Description: An automated chart audit is obtained for each patient scheduled to attend the clinic, based on protocol‐driven algorithms, and prints a report for the physician. 80% of suggestions are recommendations concerning general medicine and preventive care, and 20% are for specific speciality problems. Typology: patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: NOT CLEAR; reference: NOT CLEAR; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "Patients were assigned through the use of stratified random sample" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | "a medical research assistant interviewed each patient for every visit during the study; all study data were then coded and entered into the automated medical record system" |
| Incomplete outcome data (attrition bias) | High risk | 305 patients entered study; 52 dropped out; 133 with first 12 mo data (43.6%) |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Unclear risk | not reported |
| Adequate protection against contamination? | High risk | patients randomized |
| Methods | Cluster‐randomized trial, unit of allocation: clinic session (n = 32) | |
| Participants | Patients: percentage women: not clear, mean age: not clear; n = 6045 Professionals: faculty, interns and residents, n = 135 Setting: outpatient (general medicine clinic of Wishard Memorial Hospital, Indianapolis, USA) | |
| Interventions | Control (comparison 1): usual care Intervention (comparison 1): physician reminder Active control (comparison 2): feedback (delayed reminder) Intervention (comparison 2): physician reminder; feedback (delayed reminder) Duration of intervention: 7 months | |
| Outcomes | Quality of care: physician compliance with 13 protocols: FOBT, pneumococcal vaccination, antacids, TB skin testing, beta‐blockers, nitrates, anti‐depressants, calcium supplements, pap smear, mammography, metronidazole, digitalis, salicylates (median) | |
| Clinical area and targeted activity | Preventive care (multiple: test ordering, vaccination, prescribing) | |
| Reminder | Description: The reminders were generated the night before scheduled appointments and were placed in patients' clinical charts. Each reminder identified the patient and listed the suggested preventive care along with data from the computer record that made the patient eligible for the action, along with supporting references (same system as McDonald 1984). Typology: Patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: NO; reference: YES; at point‐of‐care: YES | |
| Notes | Data extracted from graphics. Author not contacted as publication date > 10 years. | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | "we randomized house staff by their clinic session" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Low risk | computerized data |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Low risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Unclear risk | not reported |
| Adequate protection against contamination? | Low risk | allocation by clinic session |
| Methods | Cluster non‐randomized trial, unit of allocation: clinic team (n = 5) | |
| Participants | Patients: with multiple chronic diseases, ≥ 20 yr, women: 74%, mean age: 60; n = 253 (n = 150 in included study arms) Professionals: junior and senior residents (n = 39) in 5 clinic teams (n = 25 in included study arms) Setting: outpatient (general medicine clinic, USA) | |
| Interventions | Active control: patient‐mediated intervention (questionnaire given to physician); distribution of educational material to patient Intervention: physician reminder; patient‐mediated intervention (questionnaire given to physician); distribution of educational material to patient Duration of intervention: 6 months | |
| Outcomes | Quality of care: adherence rates to recommendations: FOBT, rectal exam, mammography, pap smear, breast exam (median) | |
| Clinical area and targeted activity | Preventive care (test ordering) | |
| Reminder | Description: The computer record/reminder system used age‐ and sex‐standardized criteria for preventive care to generate patient‐specific reminder that was printed at the bottom of each patient's visit record sheet. Physicians were instructed to write in the date on which each listed service was last completed (C) or requested (R). On subsequent visits, the computer printed the updated information on the visit record; an asterisk in the "due " column indicated that it was time to repeat the service. Typology: Patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | Additional study intervention excluded from analyses: distribution of educational material to patient | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | High risk | allocation by day |
| Allocation concealment (selection bias) | High risk | allocation by day |
| Blinding of outcome assessment (detection bias) | High risk | "chart abstractors were not blind to the clinic groups of the patients in the audit" |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | High risk | tetanus vaccination rate not shown, as performance rate remained at less than 10% despite the interventions |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | High risk | "extremely poor performance of mammography by the residents in the computer/questionnaire group" |
| Characteristics at baseline similar? | Low risk | "no significant differences were found among groups" |
| Adequate protection against contamination? | Low risk | clinic teams allocated, on different days |
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: women: 49.3% , mean age: 7, n = 1619 randomized (1611 analyzed) Professionals: clinical officer, physician, nurse (n = 30) Setting: outpatient (pediatric HIV clinic, Kenya) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 5 months | |
| Outcomes | Quality of care: completion of overdue clinical tasks (primary); mean time (days) from when a patient had an overdue clinical task to the completion of the task | |
| Clinical area and targeted activity | HIV care (multiple: prescription, test ordering, referral) | |
| Reminder | The patient‐specific clinical summary, tailored for pediatric care, displayed selected information from the patient’s EHR to provide a quick reference to the most relevant data needed by clinicians. The module also contained CDSS functionality that appended patient‐ specific care reminders (overdue tests and treatments) to the bottom of the clinical summary. All summaries were attached to the relevant patients’ paper charts for clinicians to review during a patient’s clinic visit. Clinicians were asked to document their response to each reminder. No more than 5 reminders were displayed for each patient per visit. Typology: patient‐specific: YES; space for response: YES; explicit advice: YES; explanation: YES; reference: YES; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "All patients, both HIV‐exposed and HIV‐infected, previously enrolled at the study site were randomly assigned to either the intervention or control group in a 1:1 ratio by using a 4‐block randomization scheme." |
| Allocation concealment (selection bias) | Unclear risk | method of concealment not described |
| Blinding of outcome assessment (detection bias) | High risk | Authors don't mention blinding of outcome assessors. |
| Incomplete outcome data (attrition bias) | Unclear risk | 8 patients were excluded from the analyses because they were seen by a clinician involved in the study. It is unclear if all the charts have been reviewed and if no data is missing. |
| Selective reporting (reporting bias) | Low risk | Primary outcome stated in the trial registry matches the primary outcome in the published results. |
| Other bias | Unclear risk | not clear. |
| Outcomes at baseline similar? | Unclear risk | no baseline outcome measures |
| Characteristics at baseline similar? | Low risk | "There were no statistically significant differences in the demographic characteristics between the control and intervention groups." |
| Adequate protection against contamination? | High risk | Clinicians received reminders for some patients but not others. "We randomized by patient in‐ stead of by clinician, because patients typically saw whichever clinician was first available at the time of their visit, and it was not possible to tell in advance which patient a clinician would see. We understood that this could sensitize clinicians to order the indicated care for control patients, which might bias our study against finding a significant effect for the reminders." |
| Methods | Randomized trial, unit of allocation: patient | |
| Participants | Patients: women: 50% , mean age: 69, n = 396 Professionals: physicians, n = not clear Setting: inpatient (Hospital, Utah, USA) | |
| Interventions | Control: usual care Intervention: physician reminder Duration of intervention: 3 months | |
| Outcomes | Quality of care: compliance rate (primary) | |
| Clinical area and targeted activity | Digoxin intoxication (prescribing) | |
| Reminder | Description: Each night, the computer program activates the alert modules for all patients in the hospital. Alert message are formatted into a "digoxin alert report" that is sent out to a line printer in the nursing division nearest the patient. This report is placed in the patient chart by nursing personnel. Typology: Patient‐specific: YES; space for response: NO, explicit advice: YES; explanation: YES; reference: NO; at point‐of‐care: YES | |
| Notes | ||
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "a random number generator was used to assign patients to an alert or nonalert group" |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Low risk | "a blinded medical record review was carried out in accordance with a protocol aimed at identifying physician actions with possible relation to the digoxin alerts" |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | not clear |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Unclear risk | not reported |
| Characteristics at baseline similar? | Low risk | "there was no statistically significant difference between groups for sex, age, and medical/surgical service" |
| Adequate protection against contamination? | High risk | patients randomized |
| Methods | Cluster‐randomized trial (2 x 2 factorial design), unit of allocation: half‐day session (n = 40) | |
| Participants | Patients: women: 67%; mean age: 59; 4,138 patients (n = 2026 in comparison 1; n = 2112 in comparison 2) Professionals: 345 providers (faculty, residents) (approx. 8 per session) Setting: outpatient (2 clinics with 20 half‐day sessions each in Atlanta, USA) | |
| Interventions | Active control (comparison 1): educational meetings; distribution of educational material to physicians Intervention (comparison 1): physician reminder; educational meetings; distribution of educational material to physicians Active control (comparison 2): educational meetings; distribution of educational material to physicians; audit and feedback Intervention (comparison 2): physician reminder ;educational meetings; distribution of educational material to physicians; audit and feedback Duration of intervention: 3 years | |
| Outcomes | Quality of care: providers who did enough (at 36 months) Patient outcomes: glycated hemoglobin level (primary); systolic blood pressure; LDL cholesterol; patients with glycated hemoglobin < 7%; patients with systolic blood pressure < 130 mmHg (no percentage data); patients with LDL cholesterol < 100 mg/dL (no percentage data) | |
| Clinical area and targeted activity | Diabetes (general management) | |
| Reminder | Description: The reminder includes a flow sheet showing clinically relevant parameters during the present visit and several previous visits, together with specific recommendations for management. The reminders document the course of critical values, provide notice when evaluations are due, include individualized recommendations for modifications in therapy. The patient‐specific reminder is printed out and attached to the front of the chart each time a patient with diabetes presents for a visit; a specific reminder sheet will be generated for each diabetic patient to be seen each week. Typology: Patient‐specific: YES; space for response: NO; explicit advice: YES; explanation: NO; reference: NO; at point‐of‐care: YES | |
| Notes | Process data provided by authors. | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Unclear risk | half‐day medical clinic sessions randomized; randomization not explicitly described |
| Allocation concealment (selection bias) | Unclear risk | not described |
| Blinding of outcome assessment (detection bias) | Unclear risk | not clear |
| Incomplete outcome data (attrition bias) | Unclear risk | not clear |
| Selective reporting (reporting bias) | Unclear risk | In published design paper, outcomes to be reported: level of glycated hemoglobin, blood pressure and lipid levels. Intensification of therapy, eye exam and foot exam: only results on intensification reported. |
| Other bias | Unclear risk | not clear |
| Outcomes at baseline similar? | Low risk | there were no differences in compliance rates among the different intervention arms at baseline (data sent by author) |
| Characteristics at baseline similar? | Low risk | there were no significant differences among the patients assigned to residents in the different intervention arms |
| Adequate protection against contamination? | Low risk | half‐day medical clinic sessions randomized |
ACE: angiotensin converting enzyme, ARB: angiotensin receptor blocker, ED: emergency department, FOBT: fecal occult blood test, GEE: generalized estimating equation, HIV: human immunodeficiency virus, HMO: health maintenance organization, ICU: intensive care unit, LDL: low‐density lipoprotein, TB: tuberculosis, VA: Veterans Affairs
Characteristics of excluded studies [ordered by study ID]
Ir a:
| Study | Reason for exclusion |
| Reminders delivered onscreen | |
| Ineligible comparison group or inappropriate control (assistant‐initiated reminder) | |
| Reminders delivered onscreen | |
| Ineligible comparison group or inappropriate control | |
| Ineligible comparison or inappropriate control | |
| Ineligible comparison or inappropriate control | |
| Ineligible comparison or inappropriate control | |
| Reminders delivered onscreen | |
| Reminders delivered onscreen | |
| Reminders delivered onscreen | |
| Ineligible comparison or inappropriate control | |
| Ineligible comparison or inappropriate control | |
| Ineligible comparison or inappropriate control | |
| Not computer‐generated | |
| Ineligible comparison or inappropriate control | |
| Ineligible comparison or inappropriate control | |
| Reminders delivered onscreen | |
| Not computer‐generated | |
| Reminders delivered onscreen | |
| Not a reminder | |
| Reminders delivered onscreen | |
| Ineligible comparison or inappropriate control | |
| Reminders delivered onscreen | |
| Ineligible comparison or inappropriate control | |
| Reminders delivered onscreen | |
| Ineligible comparison or inappropriate control | |
| Ineligible comparison or inappropriate control | |
| Reminders delivered onscreen | |
| Reminders delivered onscreen | |
| Mixed reminder | |
| Ineligible comparison or inappropriate control | |
| Ineligible comparison or inappropriate control | |
| Mixed reminder | |
| Ineligible comparison or inappropriate control | |
| Not a reminder | |
| Ineligible comparison or inappropriate control | |
| Ineligible comparison or inappropriate control | |
| Not original study | |
| Reminders delivered onscreen | |
| Not original study | |
| Reminders delivered onscreen | |
| Not computer‐generated | |
| Not computer‐generated | |
| Ineligible comparison or inappropriate control | |
| Ineligible comparison or inappropriate control | |
| Reminders delivered onscreen | |
| Reminders delivered onscreen | |
| Ineligible comparison or inappropriate control | |
| Mixed reminder | |
| Not computer‐generated | |
| Reminders delivered onscreen | |
| Ineligible comparison or inappropriate control |
Study flow diagram
*Ineligible comparison or inappropriate control: e.g. physician reminder combined with another intervention vs usual care, physician reminder with a specific feature vs physician reminder without it, physician reminder vs another intervention
$Not a provider reminder: e.g. audit and feedback, changes in medical records system, expert system for estimating diagnosis/risk/dosage, patient‐mediated intervention
Methodological quality summary: review authors' judgements about each methodological quality item for each included study.
Methodological quality graph: review authors' judgements about each methodological quality item presented as percentages across all included studies.
Absolute improvement of quality of care by study, using the primary outcome defined by authors (represented by a red dot), and median improvement by study, using the median outcome of all reported quality of care outcomes (represented by a blue square (the median) and blue line (interquartile range))
Median effect and interquartile range (IQR) across comparisons by reminder feature (P values reflect Mann–Whitney test)
Median effect and interquartile range (IQR) across comparisons by study feature (*Kruskall–Wallis test; other P values reflect Mann–Whitney test)
| Computer‐generated reminders delivered on paper to healthcare professionals, alone or in addition to co‐intervention(s), compared with usual care or the co‐intervention(s) without the reminder component | ||||
| Patient or population: Healthcare professionals Settings: Outpatient care in Canada, France, Israel, Kenya and USA Intervention: Reminders automatically generated through a computerized system (computer‐generated) and delivered on paper to healthcare professionals, alone or in addition to one or more co‐interventions, aimed at enhancing compliance with preventive guidelines (e.g. cancer screening tests, vaccination) or disease management guidelines for acute or chronic conditions (e.g. annual follow‐ups, laboratory tests, medication adjustment, counseling) Comparison: Usual care or co‐intervention(s) without reminder component | ||||
| Outcomes | Median improvement | Number of studies (comparisons) | Certainty of the evidence | Comments |
| Quality of care | Pooling data across the 40 comparisons, the median improvement in quality of care associated with the reminder intervention was 6.8% (IQR 3.8% to 17.5%). | 34 studies (40 comparisons) | ⊕⊕⊕⊝ | Quality of care was measured by various rates: e.g. test ordering rates, vaccination rates, follow‐up rates, prescription rates, overall compliance rate. |
| Patient outcomes | Not estimable | 6 studies (7 comparisons) | ⊕⊝⊝⊝ | No measurable effect on i) blood pressure, glycated hemoglobin and cholesterol levels, ii) reaching blood pressure, glycated hemoglobin and cholesterol targets, and iii) mortality. |
| Adverse effects | Not reported | ‐ | ‐ | None of the included studies reported outcomes related to harms or adverse effects of reminders. |
| IQR: interquartile range | ||||
| GRADE Working Group grades of evidence | ||||
| 1 We downgraded the level of certainty of the evidence from high to moderate because of methodological limitations in the included studies and possible publication bias. We did not find other serious limitations in the other factors (indirectness of evidence, inconsistency of results, and imprecision of results). 2 We downgraded the level of certainty of the evidence from high to very low because of methodological limitations in the included studies, imprecision of results (wide confidence intervals) and inconsistency of the results. | ||||
| Computer‐generated reminders delivered on paper to healthcare professionals alone (single‐component intervention) compared with usual care | ||||
| Patient or population: Healthcare professionals Settings: Outpatient care in Canada, France, Israel, Kenya and USA Intervention: Computer‐generated reminders delivered on paper alone (single‐component intervention) Comparison: Usual care | ||||
| Outcomes | Median improvement | Number of studies (comparisons) | Certainty of the evidence | Comments |
| Quality of care | Pooling data across the 27 comparisons, the median improvement in quality of care associated with the reminder intervention was 11.0% (IQR 5.4% to 20.0%) | 27 studies (27 comparisons) | ⊕⊕⊕⊝ | Quality of care was measured by various rates: e.g. test ordering rates, vaccination rates, follow‐up rates, prescription rates, overall compliance rate. |
| Patient outcomes | Not estimable | 4 studies (4 comparisons ) | ⊕⊝⊝⊝ | No measurable effect on i) blood pressure, glycated hemoglobin and cholesterol levels, ii) reaching blood pressure, glycated hemoglobin and cholesterol targets, and iii) mortality. |
| Adverse effects | Not reported | ‐ | ‐ | None of the included studies reported outcomes related to harms or adverse effects of reminders. |
| IQR: interquartile range | ||||
| GRADE Working Group grades of evidence | ||||
| 1 We downgraded the level of certainty of the evidence from high to moderate because of methodological limitations in the included studies and possible publication bias. We did not find other serious limitations in the other factors (indirectness of evidence, inconsistency of results, and imprecision of results). 2 We downgraded the level of certainty of the evidence from high to very low because of methodological limitations in the included studies, imprecision of results (wide confidence intervals) and inconsistency of the results. | ||||
| Computer‐generated reminders delivered on paper to healthcare professionals in addition to one or more co‐interventions (multi‐component intervention) compared with the co‐intervention(s) without the reminder component | ||||
| Patient or population: Healthcare professionals Settings: Outpatient care in Canada and USA Intervention: Computer‐generated reminders delivered on paper in addition to one or more co‐interventions (multi‐component intervention) Comparison: Co‐intervention(s) without the reminder component | ||||
| Outcomes | Median improvement (interquartile range) | Number of studies (comparisons) | Certainty of the evidence | Comments |
| Quality of care | Pooling data across the 13 comparisons, the median improvement in quality of care associated with the reminder intervention was 4.0% (3.0% to 6.0%) | 11 studies (13 comparisons) | ⊕⊕⊕⊝ | Quality of care was measured by various rates: e.g. test ordering rates, vaccination rates, follow‐up rates, prescription rates, overall compliance rate. |
| Patient outcomes | Not estimable | 2 studies (3 comparisons) | ⊕⊝⊝⊝ | No measurable effect on i) blood pressure, glycated hemoglobin and cholesterol levels, ii) reaching blood pressure, glycated hemoglobin and cholesterol targets, and iii) mortality. |
| Adverse effects | Not reported | ‐ | ‐ | None of the included studies reported outcomes related to harms or adverse effects of reminders. |
| IQR: interquartile range | ||||
| GRADE Working Group grades of evidence | ||||
| 1 We downgraded the level of certainty of the evidence from high to moderate because of methodological limitations in the included studies and possible publication bias. We did not find other serious limitations in the other factors (indirectness of evidence, inconsistency of results, and imprecision of results). 2 We downgraded the level of certainty of the evidence from high to very low because of methodological limitations in the included studies and imprecision of results (wide confidence intervals). | ||||
| Study ID | Primary outcome | Other outcomes (n) | Absolute improvement ‐ using primary outcome | Median absolute improvement ‐ using other outcomes (interquartile range) |
|
| percentage of eligible patients with: blood pressure values on record, follow‐up (2) |
| 38.8% (18.4% to 59.1%) | |
| overall compliance rate with preventive care recommendations | percentage of eligible patients with: dental check, ocular pressure check, FOBT, flu vacc, pneumo vacc, tetanus vacc, mammography, pap smear (8) | 4.7% | 5.8% (3.0% to 10.2%) | |
| percentage of eligible patients with pap smear |
| 7.6% |
| |
| percentage of eligible patients with mammography |
| 6.0% |
| |
| percentage of eligible patients with mammography |
| 4.0% |
| |
| percentage of eligible patients with pap smear |
| 1.0% |
| |
| percentage of eligible patients with pap smear |
| 3.0% |
| |
| percentage of eligible patients with mammography |
| 7.1% |
| |
| percentage of eligible patients with flu vacc |
| 20.7% |
| |
|
| percentage of eligible patients with: discussion of directives, completion of directives (2) |
| 9.2% (6.1% to 12.3%) | |
| percentage of patients with adequate monitoring, percentage of eligible patients with initiation or up‐titration of statin therapy, percentage of eligible patients with up‐titration (3) | 5.4% (1.2% to 6.1%) | |||
| percentage of eligible patients with ACE inhibitor prescription |
| 11.5% |
| |
| percentage of eligible patients with any β‐blocker prescription | percentage of eligible patients with recommended β‐blocker prescription | 7.9% | 6.7% (5.4% to 7.9%) | |
| percentage of eligible patients with advance directives | percentage of eligible patients with: discussion or completion of directives, completion of healthcare proxy, completion of living will (3) | ‐0.2% | ‐0.3% (‐0.9% to ‐0.2%) | |
| compliance rate with prescription reminders (start a new drug) (denominator: reminders) |
| 7.0% |
| |
| adherence to colorectal cancer screening | 1.7% | |||
| overall physician compliance rate | physician compliance rate with: foot exam, physical exam, glycated hemoglobin, urine protein determination, cholesterol level, eye exam, flu vacc, pneumo vacc (8) | 16.4% | 22.7% (11.0% to 28.4%) | |
| overall compliance rate with prescription reminders | percentage of eligible patients with: ACE inhibitor or ARB therapy prescription, statins prescription | 6.0% | 9.0% (0.0% to 18.0%) | |
|
| physician compliance rate with: glycated hemoglobin, fasting blood glucose, home‐monitored blood glucose, diet clinic referral, oral hypoglycemic agents (5) |
| 4.0% (4.0% to 5.0%) | |
| overall compliance rate with prescription reminders | percentage of eligible patients with: statins, standardized statin dose, another lipid‐lowering drug, acetylsalicylic acid, acetylsalicylic acid or thienopyridine, ACE inhibitor, ACE inhibitor or ARB, β‐blocker, triple therapy (8) | 6.6% | 0.5% (‐0.4% to 2.2%) | |
| overall compliance rate with prescription reminders (denominator: reminders) | compliance with: observing a physical finding or inquiring about a symptom, ordering a diagnostic study, changing or initiating a therapeutic regimen (3) | 28.9% | 24.7% (21.1% to 38.8%) | |
| overall compliance rate with reminders (denominator: reminders) | percentage of patients with: test order, therapeutic change (2) | 23.5% | 20.3% (14.9% to 25.7%) | |
| overall compliance rate with reminders (denominator: reminders) |
| 18.6% |
| |
| overall compliance rate with reminders | percentage of patients with: FOBT, pap smear, chest roentgenogram, pneumo vacc, tuberculosis skin test, serum potassium, mammogram, flu vacc, diet, digitalis, antacids, β‐blockers (12) | 20.0% | 13.0% (10.5% to 24.5%) | |
|
| physician compliance rate with: FOBT, rectal exam, sigmoidoscopy, pap smear, pelvic exam, breast exam, mammography (7) |
| 23.0% (20.0% to 33.0%) | |
|
| physician compliance rate with: breast exam, mammography (2) |
| 23.2% (20.0% to 26.5%) | |
|
| percentage of patients with: blood group and type, syphilis serology, prenatal counseling, pregnancy diet counseling, sickle cell preparation (5) |
| 0.1% (‐1.9% to 2.0%) | |
| overall physician compliance rate with reminders |
| 3.9% |
| |
|
| percentage of eligible patients with: FOBT, mammography, tetanus vacc, cholesterol, pap smear (5) |
| 4.4% (3.9% to 6.9%) | |
|
| percentage of eligible patients with: FOBT, mammography, tetanus vacc, cholesterol, pap smear (5) |
| 6.1% (3.9% to 7.0%) | |
| overall compliance rate | percentage of eligible patients with: flu vacc, tetanus vacc, BP reading, pap smear (4) | 19.2% | 11.4% (6.0% to 16.4%) | |
| percentage of eligible patients with prescription change |
| 11.0% |
| |
| compliance rate with reminders |
| 12.9% |
| |
|
| physician compliance rate with: FOBT, pneumo vacc, antacids, TB skin testing, β‐blockers, nitrates, anti‐depressants, calcium supplements, pap smear, mammography, metronidazole, digitalis, salicylates (13) |
| 1.5% (0.5% to 11.0%) | |
|
| physician compliance rate with: FOBT, pneumo vacc, antacids, TB skin testing, β‐blockers, nitrates, anti‐depressants, calcium supplements, pap smear, mammography, metronidazole, digitalis, salicylates (13) |
| 1.0% (‐0.5% to 2.0%) | |
|
| physician compliance rate with: FOBT, rectal exam, pap smear, breast exam, mammography (5) |
| 3.6% (‐5.8% to 10.1%) | |
| completion of overdue clinical tasks (denominator: reminders) | completion of overdue clinical task for: ordering chest x‐ray, ordering 18‐mo human immunodeficiency virus enzyme‐linked immunosorbent assay, ordering other laboratory tests, beginning antiretroviral therapy, referring to nutritional support (5) | 50.0% | 39.0% (26.0% to 54.0%) | |
| compliance rate with reminders (denominator: reminders) |
| 12.0% |
| |
| physician compliance rate |
| 0.2% |
| |
| physician compliance rate |
| 0.7% |
| |
| ACE: angiotensin‐converting enzyme, ARB: angiotensin II receptor blockers, BP: blood pressure, flu: influenza, FOBT: fecal occult blood test, pneumo: pneumococcal, TB: tuberculosis, Vacc: vaccination | ||||
| Median improvement (interquartile range) | |||
| Using primary (or median) outcome | Using largest outcome | Using smallest outcome | |
| All (n = 40) | 6.8% (3.8% to 17.5%) | 12.0% (6.1% to 20.2%) | 4.0% (0.5% to 11.3%) |
| Reminders alone (n = 27) | 11.0% (5.4% to 20.0%) | 12.3% (7.0% to 33.5%) | 6.1% (1.2% to 12.9%) |
| Reminders with co‐intervention(s) (n = 13) | 4.0% (3.0% to 6.0%) | 9.8% (3.9% to 12.5%) | 0.7% (‐1.9% to 3.6%) |
| Study ID | Patient outcome: percentage difference between groups at follow‐up | Patient outcome: mean difference between groups at follow‐up |
| Percentage of patients with BP<100 or receiving treatment at 12 mo: 18.1% | ||
| Event‐free survival: ‐2.1% | LDL level: ‐2.4 mg/dL | |
| Mortality: hazard ratio: 0.98 (95% CI: 0.78 to 1.23) | Diastolic BP: 0 Systolic BP: 0 | |
| Mortality: 1% | ||
| Diastolic BP: ‐4 Systolic BP: 0 | ||
| Percentage of patients with Hba1c<7.0%: OR: 0.98 (95% CI: 0.86 to 1.12) Percentage of patients with systolic BP<130: OR: 1.04 (95% CI: 0.94 to 1.16) Percentage of patients with LDL<100: OR 0.92 (95% CI: 0.79 to 1.08) | Hba1c: 0.1 Systolic BP: ‐1.2 LDL level: 2.5 mg/dL | |
| Percentage of patients with Hba1c<7.0%: OR: 0.99 (95% CI: 0.82 to 1.19) Percentage of patients with systolic BP<130: OR: 0.92 (95% CI: 0.79 to 1.06) Percentage of patients with LDL<100: OR 1.05 (95% CI: 0.84 to 1.31) | Hba1c: 0.4 Systolic BP: 0.8 LDL level: 3.0 mg/dL | |
| BP: blood pressure, Hba1c: glycated hemoglobin, LDL: low‐density lipoprotein, mo: months | ||
