Table 2 Characteristics of Included Studies Comparing Alternative Surveillance Systems

Seaman & Lammers, 1991

433 patients with lacerations that were previously sutured, and had returned to the same county hospital for wound evaluation or suture removal.

i) Patient Interview by medical practitioner Compared with Gold Standard

ii) Direct examination of wound by physician assistant or nurse.

Unclear. Patients had to return in 2–3 days for wound check and then at 5 to 14 days for suture removal

64.5% = 433 183 (27.2%) excluded due to incomplete forms 3 (0.4%) excluded because the definition of infection by the examiner was different to that defined for the study. 52 (7.7%) Patient had prior involvement in the study

4.8%

Many methodological details not reported in this study.

Sands et al 1996

5042 patients who underwent 5572 non obstetric procedures.

i) Computerised search of electronic records with codes indicative of SSIs, which were then confirmed by record review by ICPs. Compared with

ii) Patient questionnaire and;

iii) Surgeon questionnaire

i) Records were searched 30 days post-operatively.

ii) 25^th-32^nd postoperative day

iii) 4–8 weeks post surgery.

i) 100%

ii) 33.4%

iii)79%

2%

Gold standard in this study comprised of wounds detected by any method and confirmed by case note review.

Sands et al, 1999

3636 patients undergoing 4086 non-obstetric surgical procedures.

i) Computerised search of electronic records with codes indicative of SSIs, which were then confirmed by record review by ICPs. Compared with

ii) Model created from above data

i) 30 days post discharge

ii) Unclear but uses above data.

i) 100%

ii) Unclear

2.3%

Study was a substudy of Sands et al 1996, using a subgroup of the study population from and same gold standard as previous study.

Mitchell et al, 1999

1360 patients having undergone procedures chosen to represent the major elective procedures performed by cardiothoracic, vascular, abdominal, orthopaedic and gynaecological surgeons at the hospital. They included clean and 'contaminated' procedures for which the expected postoperative stay exceeded 5 days.

i) Patient questionnaire Compared with

ii) Surgeon questionnaire

i) 28 days post-operatively.

ii) At the time of postoperative review (exact time not stated)

i) 57.5%

ii) 50.0%

6.91%

Gold standard unclear as some wounds that were detected by patients but not surgeons were reclassified as SSIs.

Whitby et al, 2002

343 patients chosen as a purposeful choice of high proportion of 'contaminated' and 'dirty' procedures.

i) Patient questionnaire Compared with Gold standard

ii) Nurse diagnosis

iii) Surgeon and Infection department physician/microbiologist assessed images of the wound

i) Questionnaire completed at 4 and 6 weeks

ii) Weekly nurse visit with visual inspection of the wound, recording of symptoms from week one to six post-discharge

iii) Not stated when the photos were taken that were assessed by these personnel.

i) 27 (7.8%) Inegible, unable, unwilling to participate. 225/316 (71.2%) completed questionnaire week 4. 190/316 (60.1%) completed questionnaire at week 6.

ii) 290/316 (91.8%) followed up weekly

iii) Not stated.

16.6%

High levels of disagreement between different health professionals were observed in this study suggesting that when one group can only make the assessment of wound infection based on photographic evidence this may not lead to accurate diagnoses.

Martini et al, 2000

1664 patients. The patient sample consisted of all surgical patients treated between May 1995 to December 1996.

i) Patient survey Compared with Gold standard of

ii) Evaluation of hospital documents and treating consultants records

i) 3 months post-discharge

ii) Not stated

i) 1535 (92%)

ii) Not stated, however no mention of missing data.

1.1%

Unclear whether patients may have sought treatment for their infected wounds elsewhere and whether the timing of the survey to patients could have led to recall bias.

Sands et al, 2003

1352 Coronary Artery Bypass Grafts (CABG) procedures performed between June 1997 and March 2003

i) Administrative claims data and pharmacy dispensing records. Compared with

ii) Examination of medical records for signs of infection using CDC criteria for wound infection by trained abstractor, further confirmed by physician epidemiologist, for a subset of patients that were at a higher probability of infection. N = 388 patients in this category

i) 180 days before surgery until 30 days post surgery.

ii) Records for each of the 30 days post-surgery.

i) Not reported

ii) 328 (85%) post-discharge information.

50.9%

Gold standard in this study comprised of wounds detected by any method and confirmed by case note review.