CHARACTERISTICS OF STUDIES

Characteristics of included studies [ordered by study ID]
Cavallini 2004
Methods	Study design: randomized clinical trial Number of study centers: 1 (University of Modena and Reggio Emilia) Number randomized: 1276 (sample size calculations based on risk of adverse events) Study follow-up: 1 month postsurgery
Participants	Country: Italy Age: not reported Gender: included men and women Inclusion criteria: patients admitted to the day surgery section at the Institute of Ophthalmology for outpatient cataract surgery under local anesthesia Exclusion criteria: ongoing treatment with anticoagulants and subcutaneous insulin therapy
Interventions	Intervention: physician review of preoperative testing, defined as routine medical tests and electrocardiograms (n = 638) Comparison: no physician review of preoperative testing, test results kept in sealed envelopes (n = 638)
Outcomes	Primary outcome: ocular adverse events, including intraoperative or postoperative adverse events Secondary outcomes: systemic adverse events defined as intra- or postoperative occurrence of acute respiratory, cardio-circulatory, or neuropsychiatric disease; or decompensation in analogous, established chronic disease
Notes	Study date: 1 October 2002 to 30 November 2003 Publication language: English
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomization list generated by Randomization Center, which was separate from the study center
Allocation concealment (selection bias)	Low risk	Medical staff at study center called into Randomization Center for participant allocation after patients were enrolled in study
Masking (performance bias) Were the participants masked to the treatment group?	Low risk	Participants were informed of the aims and methods for the study at enrollment, however all participants underwent preoperative testing
Masking (performance bias) Were the physicians performing the preoperative tests masked to the treatment group?	Unclear risk	The physicians evaluating the preoperative tests were not masked to the participants in the testing group, however they only received sealed envelopes for the participants in the non-testing group and were not informed of participants’ identities or surgery dates. It is unclear if the physician evaluating the preoperative tests was also the physician performing the surgery
Masking (detection bias) Were the primary outcome assessors masked to the treatment group?	Unclear risk	Ocular outcomes were assessed by clinical records at the time of discharge (intraoperative outcomes) and by telephone interviews 1 month after surgery (postoperative outcomes). It is unclear if the clinical records contained the treatment assignment or if the interviewers were informed of the treatment assignment
Masking (detection bias) Were the secondary outcome assessors masked to the treatment group?	Unclear risk	Systemic outcomes were assessed by clinical records at the time of discharge (intraoperative outcomes) and by telephone interviews and primary care examinations 1 month after surgery (postoperative outcomes). It is unclear if the clinical records contained the treatment assignment or if the interviewers or primary care physicians were informed of the treatment assignment
Incomplete outcome data (attrition bias) All outcomes	Low risk	No loss to follow-up is reported. Reported results are based on total number randomized
Selective reporting (reporting bias)	Low risk	Reported all ocular and systemic adverse events that occurred intraoperatively or postoperatively
Lira 2001
Methods	Study design: randomized clinical trial Number of study centers: 1 (State University of Campinas) Number randomized: 1025 (sample size calculations based on risk of adverse events) Study follow-up: up to 60 days postsurgery
Participants	Country: Brazil Age: 66.5 ±11.6 years, range 40 to 97 years (routine-testing group = 66.4 ± 11.9 years; selective-testing group = 66.7 ± 11.4 years) Gender: 547 men, 478 women (routine-testing group: men = 279, women = 233; selective-testing group: men = 268, women = 245) Inclusion criteria: people scheduled to undergo cataract surgery Exclusion criteria: less than 40 years old; undergoing surgery on the 2nd eye; were receiving general anesthesia; had a myocardial infarction within the preceding 3 months
Interventions	Intervention: routine testing with a 12-lead electrocardiogram, a complete blood count, and measurements of serum glucose (n = 512) Comparison: selective testing defined by no preoperative testing unless the participant presented with a new or worsening condition that would warrant medical testing even if no surgery was scheduled (n = 513)
Outcomes	Primary outcome: rate of complications during the perioperative period Secondary outcomes: rate of cancellation of surgery; visual acuity
Notes	Study date: 10 February 2000 to 10 January 2001 Publication languages: English and Portuguese Surgery: extra capsular extraction performed by residents under training
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomization using blocks of 4 participants
Allocation concealment (selection bias)	Unclear risk	Not described
Masking (performance bias) Were the participants masked to the treatment group?	High risk	Participants either had preoperative testing done or did not
Masking (performance bias) Were the physicians performing the preoperative tests masked to the treatment group?	High risk	Physicians performing the preoperative medical assessment knew for which participants to conduct preoperative testing
Masking (detection bias) Were the primary outcome assessors masked to the treatment group?	Low risk	Medical events and treatments were recorded by an ophthalmologist or nurse using a standardized form during surgery. The researchers reviewing the forms for classifying adverse events were masked to the treatment assignments
Masking (detection bias) Were the secondary outcome assessors masked to the treatment group?	Unclear risk	It was unclear who made the decision to cancel surgeries, or when those decisions were made
Incomplete outcome data (attrition bias) All outcomes	Low risk	Primary outcome data are presented for all participants who underwent surgery, thus for all participants at risk for complications due to cataract surgery
Selective reporting (reporting bias)	Low risk	Reported the results for adverse medical events defined in methods section using a standardized form
Schein 2000
Methods	Study design: randomized clinical trial Number of study centers: 9 Number randomized: 19,557 operations (18,189 participants) (sample size calculations based on risk of adverse events) Study follow-up: 1 week postsurgery
Participants	Country: USA and Canada Age (per operation): routine-testing group = 73 ± 8 years; no-testing group = 74 ± 8 years Gender (per operation): 7631 men; 11,926 women (routine-testing group: men = 3769, women = 6006; no-testing group: men = 3862, women = 5920) Inclusion criteria: people scheduled to undergo cataract surgery Exclusion criteria: less than 50 years old; were receiving general anesthesia; had a myocardial infarction within the preceding 3 months; had any preoperative medical testing done during the 28 days prior to enrollment; could not speak English or Spanish; 2nd eye not eligible if surgery was within 28 days of surgery in 1st randomized eye
Interventions	Intervention: routine testing with electrocardiography, complete blood count, and measurement of serum levels of electrolytes, urea nitrogen, creatinine, and glucose (operations scheduled: operations: n = 9775, participants: n = 9456; operations performed: operations: n = 9624, participants: n = 9411) Comparison: no preoperative testing unless the participant presented with a new or worsening condition that would warrant medical testing even if no surgery was scheduled (operations scheduled: operations: n = 9782, participants: n = 9445; operations performed: operations: n = 9626, participants: n = 9408)
Outcomes	Primary outcome: adverse medical events and interventions on the day of surgery and up to 7 days after surgery Secondary outcome: whether preoperative testing could have prevented the adverse event from occurring
Notes	Study date: 1 June 1995 to 30 June 1997 Publication language: English Participation rate: 94% Funding source: Agency for Health Care Policy and Research
Risk of bias
Bias	Authors’ judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Randomization was stratified according to clinical center, age (in decades), and health status reported by participants using blocks of 4
Allocation concealment (selection bias)	Low risk	Randomization was done by computer at time of enrollment.
Masking (performance bias) Were the participants masked to the treatment group?	High risk	Participants were informed of group assignment and given a letter and study brochure to present to the healthcare provider performing the preoperative assessment
Masking (performance bias) Were the physicians performing the preoperative tests masked to the treatment group?	High risk	Healthcare providers performing the preoperative tests received a letter and study brochure from the participant at the time of the preoperative assessment
Masking (detection bias) Were the primary outcome assessors masked to the treatment group?	Low risk	Medical events and treatments were recorded by an anesthesiologist or nurse anesthetist using a standardized form during surgery, and by a standardized telephone interview conducted by a study coordinator 1 week following surgery. Additional patient information was recorded by nursing staff before discharge. 2 investigators reviewed medical charts to verify adverse events, and a 3rd investigator who was masked to the treatment assignment made the final clinical judgement
Masking (detection bias) Were the secondary outcome assessors masked to the treatment group?	Low risk	2 investigators reviewed medical charts to verify adverse events, and a 3rd investigator who was masked to the treatment assignment made the final clinical judgement
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention-to-treat analysis was used. Data were 100% from day of surgery and 99.8% for 1 week after surgery
Selective reporting (reporting bias)	Low risk	Reported the results for adverse medical events defined in methods section using a standardized form and standardized telephone interview