Settings
All HH opportunities in the study were observed in Kaohsiung Veterans General Hospital (KVGH) from October 2012 through December 2013. Located in Kaohsiung City, Taiwan, KVGH is a tertiary teaching hospital with 1408 beds and more than 3600 HCWs. KVGH has regarded HH as a core element of infection control for nearly two decades and has implemented HH improvement strategies according to WHO guidelines soon after their issuance in 2009. The HH promotion campaign led to a significant decrease in the rate of healthcare-associated infection – from 3.7 to 3.1% (p = 0.002) – and reductions in overall healthcare cost and average hospital stay [16]. Direct overt observation and feedback have been routinely performed in KVGH since 2009 as described in the WHO technique manual for observers. The overall rate of HH compliance in 2011 was 73% [16].
The HH compliance of HCWs was both overtly and covertly observed in wards, intensive care units, and outpatient clinics. The study protocol was approved by the KVGH Institutional Review Board (VGHKS13-CT4–05). All covert observers signed informed consent forms. Overt observers were exempted from signing informed consent forms because overt observation of HH compliance has long been a routine infection control practice in this hospital.
Overt observers
Overt observers were recruited among HCWs at KVGH. HCWs who were interested in becoming HH compliance observers could register to receive training. A total of 38 overt observers were included in the study after receiving training and certification. Those 38 observers were assigned to specific areas in the hospital and asked to observe for at least one hour every month. The observers could freely decide their observation dates and times. The minimum length of each observation period was 20 min. The hospital administration periodically and publicly commended the overt observers for their service but provided no monetary compensation.
Covert observers
The covert observers were medical students who participated in a HH training project in 2012–2013. The project was part of a campaign to improve HH in medical students after we found that doctors had the lowest HH compliance of all HCWs. The project recruited 149 students, who, similar to their overt observer counterparts, completed the same training and underwent the same validity check. In all, 93 of these students became covert observers. Their observer duties were performed at the sites of their internship rotations in teaching hospitals across Taiwan. A previous study described detailed observation results in hospitals other than KVGH [17]. In this study, only covert observations in KVGH were used for analysis.
Training and certification
Overt and covert observers received nearly identical HH instruction from the same instructors. The course comprised lectures on HH knowledge, handrubbing and handwashing techniques, observation skills, video watching, and group discussion. The main teaching materials were the guidelines, video, and technical reference manual issued by the WHO [3, 18, 19]. After receiving the training, the observers were certified by passing a written test and demonstrating observer proficiency in five health care scenarios filmed by the infection control team. The process of HH observation via video watching also served as the means to reach inter-observer agreement. Those observers who were correct ≥80% of the time were considered qualified. Arrangements were made for both overt and covert observers to join discussion meetings every 3 months after being qualified. At those meetings, the concepts of HH observation will be reinforced, and problems about observation will be discussed to reach a consensus.
Covert observation techniques
The only difference between the courses taught to overt and covert observers was that covert observers were also taught techniques of covert observation. We designed a unique coding system to aid participants in remembering the observation results correctly and to reduce recall bias [17]. The numbers 1–5 were used to represent 5 HH indications, along with R for handrubbing, W for handwashing, and N for no HH action. For example, when a physician washed his or her hands with handrub between visiting 2 consecutive patients, observers would quietly memorize 14R. The observers were encouraged to record the codes as soon as possible after observing no more than 2–3 HH opportunities.
Observation and reporting
The locations of overt observation were assigned by the infection control team of KVGH, and the overt observers decided their observation times and durations. The covert observers were encouraged to monitor compliance during their daily routines, such as ward rounds, in order not to interfere with their internship learning or be detected by other HCWs. The observation data records were uploaded to a website that could be accessed only by qualified observers and KVGH infection control team members. The covert observers were compensated $0.30 for every HH opportunity observed and reported. Overt observers received no monetary rewards.
Statistical analysis
HH opportunities observed in the study period were analysed according to different variables using descriptive statistics. Thereafter, the overt and covert observations were matched for patient department, HCW occupation, observation location, and observation time using a 1:1 exposure and non-exposure matching method. We used the difference in the rate of HH compliance between overt and covert observers in these matched pairs as a measure of the Hawthorne effect. McNemar’s tests were used to analyse the differences between overt and covert observer data in each category. Generalized estimating equations were then used to test for differences between overt and covert observer data among categories within each variable. For example, for the variable of occupation, McNemar’s test was used to assess the differences in the compliance rates of overt and covert observation for nurses, physicians, caregivers and other HCWs, respectively; generalized estimating equations were performed to test for differences between overt and covert observer data among physicians, nurses, caregivers, and other HCWs.
Only matched pairs with the same number of HH indications regardless of observation method were included in the analysis for the variable “number of indications per hand hygiene opportunity”. Only the matched pairs with the same HH indication regardless of observation method were used to analyse the variable “hand hygiene indication”. All analyses were performed using SPSS 22.0 for Windows (IBM, Armonk, NY).