Patient-Monitored Surgical Site Infections in Karachi, Pakistan
Author(s):
Katherine Albutt; Gustaf Drevin
Background:
Background: Surgical site infections (SSI) are the most common hospital-acquired infection in low- and middle-income countries (LMICs), occurring in one in ten surgical patients and taxing already fragile health systems. The majority of SSI occur after discharge rendering prompt and accurate diagnosis and treatment challenging. Such limitations have catalyzed the development of alternate surveillance strategies, including self-monitored SSI surveillance.
Hypothesis:
We evaluated the accuracy of patient self-screening and evaluation by trained nurses, referred to as infection control monitors (ICMs), in order to develop a simple, accurate, and reproducible method of SSI detection.
Methods:
Between October 2015 and September 2017, a two-year non-inferiority study was conducted at the Indus Hospital in Karachi, Pakistan. A questionnaire designed to elicit signs and symptoms of SSI was provided to surgical patients to self-screen for infections after discharge. Results from this screening questionnaire were compared to surgeon evaluation (the gold standard for SSI diagnosis) and ICM evaluation at follow-up.
Results:
A total of 348 patients were enrolled, counseled regarding SSI signs and symptoms, and went on to complete the study. Among these patients, 18 (5.17%) developed a SSI based on surgeon evaluation. Patient self-screening had a sensitivity of 39%, specificity of 95%, positive predictive value (PPV) of 28%, negative predictive value (NPV) of 97%, and Gwet’s AC1 of 0.91 (95% CI: 0.77-0.93). The most common patient-identified symptom used to correctly identify a SSI was drainage (86%) followed by increasing surgical site pain, redness, fever, and swelling (57%, respectively). ICM screening had a sensitivity of 82%, specificity of 99%, positive predictive value (PPV) of 82%, negative predictive value (NPV) of 99%, and Gwet’s AC1 of 0.98 (95% CI: 0.91-0.99).
Conclusions:
Despite the low sensitivity observed, there was high inter-rater reliability between surgeon diagnosis of SSI and patient self-screening. The high specificity and low false positive rate suggest that patients are able to correctly identify an uninfected surgical wound. Supplementing regular post discharge follow-up with patient self-screening in a low-resource settings has the potential to increase the rate of SSI detection with minimal additional burden to the health system. There was near perfect agreement between surgeon diagnosis and ICM assessment at follow-up with high sensitivity and specificity. Trained nurses can correctly identify SSI and may be used as a proxy to surgeons for SSI detection, thereby reducing the burden on the specialized surgical workforce in LMICs.