Setting and study subjects

This study was conducted in a sample of rural hospitals in Idaho and Utah. Hospitals were recruited to participate from a larger pool of rural hospitals in both states already involved in a laboratory surveillance study for resistant pathogens[25]. Inclusion criteria for the current study were: a) hospitals meeting the Office of Management and Budget definition of rural location[26]; b) hospitals indicating the availability of alcohol-based hand gels in their facility; c) hospitals with administrative support for active interventions including inpatient surveillance culturing; and d) hospitals with an infection preventionist (IP) willing to supervise HH compliance monitoring. The selected hospitals were separated by significant geographic distances in both states.

HCWs of all types in each hospital were the object of observation and monitoring of HH compliance. Observers were recruited by the IP from each hospital to conduct blinded observations of HCWs for HH compliance. During the time frames selected to determine the point-prevalence of methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci (VRE) by active surveillance culturing, those patients selected as candidates for providing surveillance cultures included any inpatient ≥18 years of age capable of giving informed consent. The University of Utah Institutional Review Board (IRB) and Western IRB (Olympia, WA) approved the study; informed consent was required for inpatients providing surveillance cultures.

Design

The structure of this study was a cluster-randomized trial[27] consisting of a common 4-month baseline measurement period followed by randomization of hospitals into “control” or “intervention” groups and a subsequent intervention time period of 5 months to examine the feasibility of implementing the research protocol. The design of this controlled clinical trial was considered pragmatic in that it addressed effectiveness or the degree of benefit of these infection control interventions in a real practice setting[28]. The study was conducted between March, 2003 and February, 2004.

HH monitoring

Observers were recruited for each hospital by the IP. IPs from each participating hospital received training from the investigative team in a group session. Each IP was provided with a training manual, a standard data collection form, and uniform definitions for data collection. A training videotape was developed which included 19 scenarios representing different types of HH opportunities. The IPs completed data collection forms while watching the videotape to test inter-rater reliability and resolve areas of disagreement. IPs subsequently trained local observers. Central research staff monitored protocol adherence of local observers during study site visits; however, it was not feasible to measure the inter-rater reliability of local observers across different hospitals.

HH observations were conducted at randomly selected times of the day between 8 am and 5 pm on weekdays in 30-minute increments following established standards for HH observations[10, 11]. Variables recorded included type of HCW, nature of patient contact and type of care, contact with the patient care environment, type of hand hygiene (hand washing with soap and water or use of alcohol hand gel), and HH before and/or after patient contact. The number of observed opportunities for hand hygiene was defined as the denominator, and the number of observed opportunities in which hand hygiene was performed served as the numerator. Percent compliance with HH recommendations was calculated. “Complete compliance” was defined as alcohol gel rub or hand washing before and after patient/environmental contact; “any compliance” was defined as alcohol gel rub or hand washing before or after patient/environmental contact. The duration or efficacy of HH technique was not measured. HH compliance was monitored in all hospitals during the baseline and intervention time periods.

Intervention components

After the 4-month baseline time period, a hospital-wide campaign to improve compliance with HH and isolation precautions was conducted in each intervention hospital while the control group continued with their usual IC practices. Each hospital staff developed a customized individualized campaign to promote compliance with HH and isolation practices. Elements of this campaign included education sessions, ensuring availability of alcohol-based hand cleaners and personal protective equipment at all patient care areas, written materials, academic detailing, displaying posters that emphasize the importance of hand hygiene and standard precautions, and recognition and rewards programs. The IP of each hospital was instructed to organize the campaign for their hospital.

Results of active surveillance cultures were provided to the intervention hospitals concurrently and to the control hospitals at the conclusion of the study. If the clinical MRSA infection rate exceeded the pre-determined threshold (3 standard deviations above the group mean), the investigative team offered assistance with active infection control interventions to curb further transmission. The following types of additional interventions were considered: education on HH and isolation precautions; campaign to promote isolation of patients; additional surveillance cultures; and decolonization of culture-positive patients[16, 29, 30].

Active surveillance cultures

Clinical cultures obtained for clinical indications which yielded MRSA or VRE were continuously reported and analyzed by the hospitals as previously described[25]. Additionally, active surveillance culturing (nasal swabs for MRSA and rectal swabs or stool cultures for VRE or MRSA) on all hospitalized patients ≥ 18 years of age giving informed consent were conducted on four separate days during the study in both groups of hospitals to determine the prevalence of colonization. These were conducted consistently by one member of the investigative team with the assistance of the local IP or trained observer. Swabs were plated for S. aureus (blood agar or mannitol salt agar) or enterococcal species (blood agar or bile esculin azide agar). Presence of methicillin or vancomycin resistance was detected by standard methods (conventional MIC testing, oxacillin salt screening agar, or selective media containing vancomycin)[31].

Statistical methods and data analysis

The primary endpoint was the proportion of health care providers in targeted hospitals compliant with HH protocols (number of compliant observations/total number of observations). This was stratified by “complete compliance” or “any compliance” as outlined above. Secondary endpoints included the rate of clinical cases of MRSA or VRE infection calculated as the number of cases per 1000 patient days and the point prevalence of patients colonized with MRSA and VRE in all participating hospitals as determined by active surveillance culturing (% colonized patients/total patients cultured).

HH compliance data were analyzed using mixed effects logistic regression to model the probability of compliance. These models accounted for the lack of independence introduced by the individual hand hygiene opportunities being nested within hospital. The study group indicator variable, intervention relative to non-intervention, and intervention period indicator variable, post-intervention relative to pre-intervention, were included as main effect terms. The interaction term, study group × intervention period, was used to statistically test the null hypothesis that the change in probability of compliance was equal in the two study groups. To provide a clear presentation of the results, both group and hospital-specific predicted probability of compliance was estimated from the models for both the pre- and post-intervention periods, and then subtracted to measure the absolute change in compliance. This allowed communication of the results in the more readily understood terms of absolute percent change rather than odds ratios.

Study hospitals

Ten rural hospitals in diverse geographic locations in Idaho and Utah were recruited to participate in this study performed in 2003. Five hospitals were each randomized to the intervention or control arms of the study. One control hospital elected to withdraw from the study shortly after the intervention time period had begun and was not included in the final analysis. The mean bed size of the intervention hospitals (62 licensed beds) was not significantly different from the control group (44 licensed beds, p = 0.62). Four of the five intervention hospitals reported that they had little awareness of the magnitude of MRSA cases or the prevalence of colonized patients in their facilities prior to initiation of this study.

Hand hygiene component

A total of 2,654 hand hygiene opportunities were observed in the five intervention hospitals and 1,873 in the four control hospitals. The two study arms were well-balanced with respect to the percent of opportunities observed across HCW occupations (Table 2). The estimated absolute change in “complete” HH adherence ranged from 7.8% to 35.5% in intervention hospitals and −6.3% to 5.9% in control hospitals. The average absolute change in intervention hospitals was 20.1% in intervention hospitals compared to −3.1% in control hospitals (p = 0.001) (Figure 1a). The estimated absolute change in “any compliance” ranged from 17.8% to 38.2% in intervention hospitals and −16.7 to 20.7% in control hospitals, with an average absolute change of 28.4% in intervention hospitals compared to 0.7% in control hospitals (p = 0.01) (Figure 1b).

Occupation	Nonintervention	Intervention
	[N = 1,873]	[N = 2,654]
Doctor (MD), n (%)	187 (10)	232 (9)
Nurse (RN)	934 (50)	1,328 (50)
Nurse assistant (CNA)	219 (12)	299 (11)
Respiratory therapist	50 (3)	72 (3)
Physical therapist	49 (3)	65 (2)
Radiation technician	87 (5)	76 (3)
Laboratory technician	82 (4)	137 (5)
Environmental services	72 (4)	93 (4)
Other	193 (10)	352 (13)

Surveillance cultures

Among the 9 participating hospitals, 290 unique surveillance cultures were performed during the total 9 month study time period. Among these were 171 nasal swabs alone (rectal swabs being refused by these patients) and 119 instances where both a nasal and rectal swabs were obtained from the same patient. The prevalence of MRSA carriage was 28 of 290 (9.7%) and of VRE carriage was 1 of 119 (0.84%). Rectal swab cultures detected three MRSA carriers not detected by nasal swab. The median number of unique positive cultures of MRSA per hospital was two (range: 0 to 11). As results of active surveillance cultures were communicated, two of the smaller hospitals initiated new programs to identify previously infected or colonized MRSA patients upon readmission. One hospital identified a local long-term care facility (LTCF) as a source of MRSA cases in their hospital and provided education resources to the LTCF.

Rural facility MRSA cluster

The results of one specific intervention are highlighted to illustrate the impact of investigator interaction with the participating facilities on IC practices. One intervention hospital with an attached long-term care unit (LTCU) had a noticeably high number of positive MRSA surveillance cultures and was the focus of targeted infection control interventions. This hospital was very small with an average daily census of 5 patients. There was a cluster of positive surveillance cultures initially noted in June 2003 (n = 5) with 4 cultures from residents of the LTCU. This number of positive cultures exceeded the predetermined threshold of 3 standard deviations above the mean for all participating hospitals. Surveillance cultures were repeated in the acute care and long-term care unit in July 2003 with 3 positive cultures noted in new patients and 1 positive culture from a patient that had been previously negative in June 2003. Interventions began in August 2003 with increased education, repeating of active surveillance cultures, selective de-colonization of all patients with new and current MRSA colonization, enhanced HH, and more aggressive application of contact precautions. Decolonization of all residents with positive surveillance cultures was performed by implementing bathing with chlorhexidine soap, application of mupirocin ointment to anterior nares, and oral antimicrobial therapy (trimethoprim/sulfamethoxazole and rifampin) according to published protocols[30]. These decolonization interventions were applied for 7 days with follow-up surveillance cultures in 4–6 weeks. Repeat surveillance cultures in November 2003 revealed no further colonized patients. There was only one patient with MRSA infection identified in the next 12 months and this was identified upon admission and not associated with healthcare transmission (Figure 2).