UBC Faculty Research and Publications

Contact among healthcare workers in the hospital setting: developing the evidence base for innovative… English, Krista M; Langley, Joanne M; McGeer, Allison; Hupert, Nathaniel; Tellier, Raymond; Henry, Bonnie; Halperin, Scott A; Johnston, Lynn; Pourbohloul, Babak Apr 17, 2018

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-12879_2018_Article_3093.pdf [ 1.9MB ]
JSON: 52383-1.0365776.json
JSON-LD: 52383-1.0365776-ld.json
RDF/XML (Pretty): 52383-1.0365776-rdf.xml
RDF/JSON: 52383-1.0365776-rdf.json
Turtle: 52383-1.0365776-turtle.txt
N-Triples: 52383-1.0365776-rdf-ntriples.txt
Original Record: 52383-1.0365776-source.json
Full Text

Full Text

RESEARCH ARTICLE Open AccessContact among healthcare workers in thehospital setting: developing the evidencebase for innovative approaches to infectioncontrolKrista M. English1, Joanne M. Langley2, Allison McGeer3, Nathaniel Hupert4, Raymond Tellier5, Bonnie Henry6,Scott A. Halperin7, Lynn Johnston8 and Babak Pourbohloul1*AbstractBackground: Nosocomial, or healthcare-associated infections (HAI), exact a high medical and financial toll on patients,healthcare workers, caretakers, and the health system. Interpersonal contact patterns play a large role in infectious diseasespread, but little is known about the relationship between health care workers’ (HCW) movements and contact patternswithin a heath care facility and HAI. Quantitatively capturing these patterns will aid in understanding the dynamics of HAIand may lead to more targeted and effective control strategies in the hospital setting.Methods: Staff at 3 urban university-based tertiary care hospitals in Canada completed a detailed questionnaireon demographics, interpersonal contacts, in-hospital movement, and infection prevention and control practices.Staff were divided into categories of administrative/support, nurses, physicians, and “Other HCWs” - a fourthdistinct category, which excludes physicians and nurses. Using quantitative network modeling tools, we constructedthe resulting HCW “co-location network” to illustrate contacts among different occupations and with locations inhospital settings.Results: Among 3048 respondents (response rate 38%) an average of 3.79, 3.69 and 3.88 floors were visited by eachHCW each week in the 3 hospitals, with a standard deviation of 2.63, 1.74 and 2.08, respectively. Physicians reportedthe highest rate of direct patient contacts (> 20 patients/day) but the lowest rate of contacts with other HCWs; nurseshad the most extended (> 20 min) periods of direct patient contact. “Other HCWs” had the most direct daily contactwith all other HCWs. Physicians also reported significantly more locations visited per week than nurses, other HCW, oradministrators; nurses visited the fewest. Public spaces such as the cafeteria had the most staff visits per week, but theleast mean hours spent per visit. Inpatient settings had significantly more HCW interactions per week than outpatientsettings.Conclusions: HCW contact patterns and spatial movement demonstrate significant heterogeneity by occupation.Control strategies that address this diversity among health care workers may be more effective than “one-strategy-fits-all” HAI prevention and control programs.Keywords: Hospital associated infections, Infection prevention and control, Contact networks* Correspondence: babak.p@ubc.ca1Institute for Resources, Environment and Sustainability, University of BritishColumbia, 2202 Main Mall, Vancouver, BC V6T 1Z4, CanadaFull list of author information is available at the end of the article© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.English et al. BMC Infectious Diseases  (2018) 18:184 https://doi.org/10.1186/s12879-018-3093-xBackgroundNosocomial, or healthcare-associated infections (HAI)are a major burden to public health and the functioningof modern healthcare systems. In Canada, more than200,000 patients acquire a HAI annually, and as a result,an estimated 8000 die [1]. Figures in the United Statesand Europe are comparable on a per-capita basis [2, 3].The 2003 severe acute respiratory syndrome (SARS) out-breaks, and more recently of Middle East respiratorysyndrome (MERS), highlight the major threat posed byHAIs, both within the hospital and for the wider commu-nity. Close contact between patients and/or healthcareworkers (HCWs), and high concentrations of medically-vulnerable populations, combined with physical movementbetween treatment areas, are factors that may facilitate HAIspread within health care institutions and the community.Current infection prevention and control (IPC) measuresfocus on proper performance of both routine practices (e.g.hand and respiratory hygiene) and additional precautions(e.g. airborne, contact and droplet precautions) by allHCWs [4, 5]. Before patient contact, HCWs determineprecautions to be taken based on their own situationalrisk assessment. However, heterogeneity of collectivecontacts among patients and HCWs are not specificallyaddressed in the current guidelines. Preliminary at-tempts to quantify mixing patterns and contact rateshave been conducted among the general population on alarge scale [6–9], or in non-healthcare settings [10–12],but rates of HCW contacts within healthcare settings arepostulated to be significantly higher and more heteroge-neous than those within the general population [13].Studies using electronic medical records to examinespatial movement throughout the hospital provide infor-mation on only a small subset of hospital interactions.These studies capture patient movement as it pertainsexplicitly to the more complex clinical services they receivebut fail to capture HCW social or casual movement, suchas visits to the cafeteria or meeting rooms, or some typesof clinical contact (e.g. a second staff member assistingwith patient mobilization or bathing, or cross-covering acolleague on break) [14–16]. Contact patterns for HCWshave been examined using radio frequency identification(RFID) tags, mote-based sensors, and direct observation[17–20]. These formats have suggested the potential for“super spreaders” in the hospital setting [20], and notabledifferences in contact patterns between occupations [19].Since these studies are currently only within a single wardor unit, they are limited in generalizability to a hospital-wide setting since they do not take contacts outside thestudy setting into account. In addition to room-levelcontacts, it is important to note the patterns of movementthroughout the hospital. This may reveal locations thatcan more readily propagate infection spread duringoutbreak scenarios.Understanding the movement and contact patternsof HCWs within hospital settings may allow for moretargeted and effective infection control interventions.To address this knowledge gap, we conducted a cross-sectional study of HCW in three major Canadian healthcare facilities to assess interpersonal contact patterns,movement throughout the facility, and demographic char-acteristics. These data can be used to develop a model thatrepresents the heterogeneous contact patterns in thehospital setting. Additional questions on IPC practiceswere included to help parameterize future models ofHAI reduction interventions.MethodsUsing architectural maps and floor plans, site-specificsurveys were created for three urban university-affiliatedtertiary care Canadian hospitals (hereafter called HospitalA, B and C). The data collection instruments were hard-copy paper booklets with information packages, containingguidelines and rationale for the study, and 1 online survey.Employees were invited to participate through personalinvitations, email and posters. Surveys were also attachedto employee paystubs on two separate occasions. Thepaper surveys were to be completed by HCWs andreturned anonymously to a centrally-located drop box.Local study staff informed participants that surveycompletion was voluntary and anonymous. This projectwas funded by the Canadian Institutes of Health Research(CIHR) and called the CONNECT I study. Ethics reviewboards at all participating universities and hospitalsapproved the project.An estimated 8100 staff working (or volunteering) inany of the three hospitals were eligible to participate(~ 4100, ~ 2400, and ~ 1600 in Hospitals A, B and C,respectively). Our pre-survey target for participationwas 1000 or 12.5%. The survey identified 19 differentHCW occupational categories including attending andresident physicians, nurses, technicians, support staff,undergraduate trainees and other hospital workers whohave patient contact. For this publication, all occupationalcategories other than physicians, nurses, and administra-tive/support staff are grouped together as a fourth maincategory called “other HCWs” (hereafter, oHCW). Thesecategories are summarized in Table 1.The surveys collected demographics, spatial movement,and patient interaction (contact) data, as well as self-reported compliance with IPC practices by both thesurvey respondent and his or her coworkers. Directpatient contact was defined as two or more individualscoming within 1 m (approximately 3 ft) of each other for2 min or more. This proximity has long been proposed asa guideline for the range of transmission of infection bylarge droplets. At the time of conducting the survey, thisproximity and duration were estimated to be necessaryEnglish et al. BMC Infectious Diseases  (2018) 18:184 Page 2 of 12but not sufficient for respiratory infection transmission (inmore recent guidance, 2 m is considered the radius forpotential transmission [21]). Indirect contact was definedas two or more individuals co-locating in the same roombut not closer than 1 m.For demographic analyses, differences between groupswere assessed using Chi-square tests and analysis ofvariance (ANOVA).Hospital floors were identified as predominantlypatient-care area (PCA), predominantly non-patient-carearea (non-PCA) and mixed (mPCA), by local study staff.Respondents reported the amount of time (in hours, orminutes) they averaged weekly in each location within theirhospital. Detailed spatial locations such as pre-admissionunit, day surgery unit, ambulatory internal medicine clinicor cafeteria, were identified in the questionnaire corre-sponding to each hospital. There were 251, 122, and97 units in Hospitals A, B and C, respectively. The fre-quency of visits and mean reported hours were quantifiedfor each location, and analysis of variance (ANOVA) wasused to compare the groups. Tukey pairwise tests wereused post-hoc to identify significant comparisons. Giventhe diversity and frequency of these small locations, it wasnecessary to aggregate the information that is simple topresent and consistent across all sites. Since this paperconcerns the structure of interpersonal HCW contacts anddoes not address the transmission dynamics of infectionspread, we group these small locations to present theresults for each actual hospital floor, as a spatial unit.Infection prevention and control practices were assessedwith questions about regular compliance with IPC precau-tions as well as through the use of three HCW-patientcontact scenarios involving a patient who is diagnosedwith a) respiratory tract infection (e.g. RSV) that is spreadby droplets; b) active pulmonary tuberculosis, who has aproductive cough; and c) varicella (chickenpox). Respon-dents were asked about the precautions they would take -such as wearing surgical mask, N95 respirator, face or eyeshield, gloves, gown, or goggles - in each of these threescenarios. Additionally, for scenario (a), they were askedto provide a response in a situation when they are within1 m (3 ft) of the patient with respiratory tract infection.Also, for scenario (c), they were asked to provide a responseassuming they had immunity to varicella (e.g., via childhoodinfection). Quantitative responses were measured on a1 to 10 scale. Charting practices regarding the accuraterecording of number of daily patient-HCW interactionswere also assessed.ResultsTwo thousand eight hundred thirteen staff completedpaper questionnaires while 235 completed electronic sur-veys. Three thousand forty-eight HCW participated (38%),which exceeded our target participation rate by three-fold.The distributions of survey participation by hospitaland occupation are summarized in Table 2. Nurses werethe occupational category with the highest aggregateresponse rate, although more administrative/supportstaff responded in Hospital A.The median age of respondents across all sites was42 years, 81% were female, and most (75%) worked in apatient-care area. More than one third (37%) of physiciansworked in other healthcare facilities in addition to thestudy hospital (Table 3).Staff visited an average of 3.79, 3.69 and 3.88 floors intheir respective healthcare facility per week, with astandard deviation of 2.63, 1.74 and 2.08. Physiciansreported the highest number of locations visited perweek, while nurses reported the lowest. The number oflocations visited varied significantly depending on jobcategory (Table 3). Results from Tukey post-hoc analysesshowed nurses visited significantly fewer locations com-pared to physicians, “other HCW” and admin/support(p = 0.002, p = 0.001, p < 0.001, respectively).Table 4 details the amount and type of contacts foreach occupational group. Physicians reported the highestnumber of direct patient contacts (> 20 patients/day) butthe lowest number of contacts with other HCWs, whilenurses had the most extended (> 20 min) periods ofTable 1 Classification of aggregate categories based onself-reported occupationsSelf-reported Occupation Category1 Central Supply Technician Admin/Support2 Housekeeping3 Receptionist4 Service Assistant5 Volunteer6 Ward Clerk7 Nurse Nurse8 Nursing Student9 Staff Physician Physician10 Postgraduate Medical Trainee11 Medical trainee12 Medical Imaging Technologist oHCW13 Patient Attendants/Sitters14 Pharmacist15 Physiotherapist/Occupational Therapist16 Respiratory Therapist17 Social Worker18 Othera19 Other Student DisciplineaaRespondents were assigned to one of the four above categories based ontheir description in the free-text field providedEnglish et al. BMC Infectious Diseases  (2018) 18:184 Page 3 of 12direct patient contact. oHCWs had the most direct dailycontact with other HCWs (Table 4).Table 4 shows the number of contacts per occupa-tional category. The first row in each pair correspondsto the number of respondents who answered questionslabeled A1,. .., D1; the second row shows the percentageof these responses that satisfied the stated criteria (e.g.,had direct contacts lasted more than 20 min).Contact network visualizationsTo provide additional insight into the aggregate statisticspresented in Tables 3 and 4, Fig. 1 illustrates HCWs’time spent on each floor at one of the participating studyhospitals. Each bar chart (row) in this figure correspondsto a separate floor in that hospital (labeled L1 – L17).Along the horizontal axis, 1512 thin bars represent 679administrative/support staff (red), 561 nurses (blue), 104physicians (cyan) and 168 oHCW (green), who respondedto the survey. The vertical axis represents time in logarith-mic scale; each bar’s height reflects the time reported bythat worker as having been spent on that floor. Thus, if aHCW reported spending a few-, up to 100 min on anysingle floor, the bar representing her/him can rise to themiddle tick on the vertical axis; if hundreds of minutes,the bar may end in the middle segment of the y-axis; andfinally, if few thousand minutes (up to a full work week),the bar may end on the upper segment of the y-axis. Also,in this figure, if a HCW spends time on more than onefloor during the week, then they are represented by non-zero bars in the bar charts corresponding to those floors(and blank space in bar charts corresponding to otherfloors). There is a great variability in terms of the reportedtime spent, during a single, multiple, or routine visit(s), oneach floor ranging from a few minutes to nearly a fullwork week (35 h/week = 2100 min/week).Based on data shown in Fig. 1, we generated avisualization of the bipartite network that captures HCWmovement within a hospital setting (Fig. 2). A bipartitenetwork shows the relationship between two distinctclasses of nodes, in this case hospital floors and HCWs.Here the array of larger yellow nodes represents differentfloors in Hospital A, while all other nodes representHCWs. An edge (black line) is drawn between a specificHCW and a location when the HCW reported visitingthat location. HCW nodes are colored based on theiroccupational category.The heterogeneity in the duration of time spent by aHCW in a spatial unit implies that the links connectinghospital floor and HCW do not have equal significancewith respect to respiratory-borne infection transmission.For low- to moderately contagious infections, the prob-ability of transmission among contacts in close-proximityis generally considered to be proportional to the durationof contact for each pair of individuals [22–24]. To accountfor the duration, each link should be weighted accordingto the length of time spent in a spatial unit; the longer theduration, the higher the weight.Incorporating weighted edges in the network results ina gravity-centered network layout shown in Fig. 2, whereedges with higher weights (“stronger” edges) and theirassociated nodes are concentrated near the core, whileedges with lower weights (“weaker” edges), and theirassociated nodes are pushed outward to the periphery ofthe network.The irregular density of edges in Fig. 2 reveals consid-erable heterogeneity in both the number and duration ofcontacts in the study hospitals. For infectious pathogenswhose probability of transmission is proportional to theduration of contact (directly between individuals, orindirectly between a person and a spatial unit), this mayTable 2 Occupational response rates for each hospital surveyedAdmin/Sup N (%) Nurses N (%) Other HCW N (%) Physicians N (%) Total RespondentsaHospital A 753 (46.3) 591 (36.4) 173 (10.6) 108 (6.6) 1625Hospital B 233 (28.7) 346 (42.7) 152 (18.7) 80 (9.9) 811Hospital C 129 (22.2) 238 (40.9) 137 (23.6) 78 (13.4) 582Total (%) 1115 (36.9) 1175 (38.9) 462 (15.3) 266 (8.8) 3018a30 non-categorized responses were excluded from this tableTable 3 Summary of location data for each occupational categoryProvided locationdata (#)Visited > 4 floorsper week (%)Visited only 1 floorper week (%)Works in patient carearea (%)Also works in anotherhealthcare facility (%)Admin/Sup 1031 29.5 16.3 42 9Nurses 1143 20.1 15.3 98 13Other HCW 452 45.6 7.96 81 15Physicians 254 47.2 10.6 92 37All Categories 2880 29.9 14.1 75 14English et al. BMC Infectious Diseases  (2018) 18:184 Page 4 of 12have a significant impact on the transmission pathwayswithin a healthcare setting. The likelihood of igniting aHAI outbreak, or being infected during such an outbreak,is higher for the nodes that are part of the central clusterthan the ones belonging to the dendritic branches in theperiphery of the network.Decomposing this network structure into its constituentoccupational categories further exposes this heterogeneity.Fig. 3 shows the underlying weighted networks of the fouroccupational categories stratified around a central imagethat is a smaller replica of the full network (i.e., Fig. 2).While most nodes corresponding to participatingTable 4 Number of contacts per occupational categoryOccupational Category→Admin/Sup Nurses Other HCW Physicians % Total %Question ↓A1) Direct contact with patients per day 1019 1159 459 262 2899A2) Direct contact with > 20 patients per day (% of A1) 261 (25.6%) 260 (22.4%) 106 (23.0%) 88 (33.6%) 715 (24.7%)B1) Direct contact with any one patient per day 1022 1158 458 262 2900B2) > 20 min of direct contact with any one patient per day (% of B1) 71 (6.9%) 781 (67.4%) 237 (51.7%) 77 (29.4%) 1166 (40.2%)C1) Indirect contact with patients per day 985 1140 449 249 2823C2) Indirect contact with > 20 patients per day (% of C1) 225 (22.8%) 274 (24.0%) 130 (29.0%) 60 (24.1%) 689 (24.4%)D1) Direct contact with HCWs/day 1012 1140 445 243 2840D2) Direct contact with > 20 HCWs/day (% of D1) 223 (22.0%) 253 (22.2%) 127 (28.5%) 34 (14.0%) 637 (22.4%)Fig. 1 Detailed data corresponding to the time spent by HCWs on each floor of one study site during a typical week. Please see the main textfor details. Different floors are labeled from L1 – L17. The bars are not sorted so that each HCW is represented on exactly same location on all 17horizontal axes. Floor labels 1–17 correspond to floor levels L1 – L17 in Fig. 1, respectivelyEnglish et al. BMC Infectious Diseases  (2018) 18:184 Page 5 of 12Administration and Nurse categories occupy the periph-eral branches of the weighted network, the majority ofphysicians are clustered in the centre (grey backgroundarea in all panels).For location analyses, both the number of visits perweek, and the mean hours spent, significantly differed bylocation type (Table 5). Public spaces had the most visitsper week but the fewest mean hours spent (0.9 h).Inpatient settings had significantly more visits per weekthan outpatient settings.The network in Fig. 2 can be divided into 3 disjointnetworks based on hospital floors’ classification as PCA,non-PCA, or mixed (Fig. 4). The sub-network for PCA(top-left panel in Fig. 4) shows 3 different patterns:nodes (outer clusters) corresponding to HCW who visitonly one floor; nodes (intermediate clusters) interactwithin two floors; and the remaining nodes (centralcore) representing individuals who visit several floors.Comparatively speaking, floors with predominantlynon-PCA areas (top right panel in Fig. 4) have higherbetween-floor traffic rate than PCA floors (top leftpanel). The highest between-floor HCW traffic occurs inmixed areas (lower panel in Fig. 4).Finally, as with Figs. 2 and 3, the sub-network corre-sponding to the PCA floors (top left panel in Fig. 4) maybe stratified into the four occupational categories (Fig. 5).All occupational categories include nodes that reportmovement between multiple PCA floors (i.e., the mostFig. 2 HCW-location bipartite network constructed from survey datacollected from one of the participating hospitals. This network isweighted by duration of visits and differentiated by type of HCW(red: administrative/support staff, 679 nodes; blue: nurses, 561 nodes;cyan: physicians, 104 nodes; and green: oHCW, 168 nodes). Ratherthan organizing all floors (larger yellow nodes) on a straight line, thepositions of these nodes are adjusted to allow for better visualizationof clustering effect among other nodesFig. 3 HCW-location bipartite network stratified by occupational category (red: administrative/support staff; blue: nurses; cyan: physicians; andgreen: oHCW)English et al. BMC Infectious Diseases  (2018) 18:184 Page 6 of 12central clusters of nodes, Table 6). This movement maycontribute to increasing the likelihood of an infectioustransmission event within PCA floors.Infection prevention and control practicesAlthough respondents reported that they believed themajority of their HCW colleagues would comply withIPC guidelines (61.5% “mostly” comply, 31.5% “partially”comply), there was wide variability in reported use ofpersonal protective equipment and only 81–87% expectedcompliance with handwashing after interacting withpatients with communicable respiratory diseases (Table 7).Additionally, most respondents believed that patientcharts would inaccurately report single or multipleHCW-patient interactions.DiscussionThe CONNECT I survey results presented here providethe most comprehensive picture of hospital-wide contactnetworks yet published. These insights provide evidenceto support the development of novel network-basedstrategies for the prevention and control of HAI. Sincethe SARS outbreaks in 2003, there has been an emergingrecognition of the complexity of hospital-based contactstructures, and that this complexity varies by occupationaltype [25]. While prior studies have focused on individualhospital wards [18, 19, 26], patient-to-patient contact [15],or simulated/hypothetical patient-to-HCW contact [27],we report on actual self-reported patterns of movementand contact of over 3000 HCW in three Canadian urbantertiary care university affiliated hospitals. The resultingfacility-specific networks identify occupational categoriesTable 5 Summary of number of visits per week and hours spent per week corresponding to each hospital in patient care (PCA),non-patient care (Non PCA), and mixed (mPCA) areasNon PCA mPCA PCAHospital 11512 RespondentsTotal number (and %) of floor visits per week 1277 (22.3%) 2725 (47.5%) 1734 (30.2%)Average time spent per floor (hrs/week) 6.5 7.5 13.0Hospital 2801 RespondentsTotal number (and %) of floor visits per week 193 (6.5%) 1262 (42.7%) 1501 (50.8%)Average time spent per floor (hrs/week) 4.9 7.7 11.7Hospital 3568 RespondentsTotal number (and %) of floor visits per week 464 (21.0%) 590 (26.7%) 1153 (52.2%)Average time spent per floor (hrs/week) 3.7 9.0 11.8Fig. 4 Three disjoint networks based on hospital floors’ classification as PCA, non-PCA, or mixed (red: administrative/support staff; blue: nurses;cyan: physicians; and green: oHCW)English et al. BMC Infectious Diseases  (2018) 18:184 Page 7 of 12and specific locations within each unique setting that havehigh and low contact rates. Such data that can be utilizedto inform targeted and efficient IPC strategies.Contact and movement patterns of HCWs varied sig-nificantly by occupation. Although more nurses reportedextended periods of direct patient contact, “other HCWs”(non-physician, non-nurse) had significantly more HCWcontact per week than any other occupational category. Inthis paper, we aggregated HCW occupations into 4 maincategories. We recognize HCW occupations such asrespiratory therapists and personal care attendants mayplay a key role in spreading micro-organisms throughphysical contact, procedures such as intubation, or patientmovement through the hospital. A higher resolutionanalysis of the survey data to address more refined ques-tions may constitute the subject of future publications.The mobility of these occupations within a hospital mayfacilitate disease propagation compared to a more local-ized (within ward) movement, such as for nurses. Mod-eling the movement of these healthcare workers in thehospital setting may provide further insight into thepropagation of diseases throughout the hospital.We found that physicians, although mobile throughoutthe hospital, have a lower length of contact with otherHCWs compared to any other occupational category,where a contact was defined as within 1 m of another in-dividual for 2 min or more. This agrees with a study onone pediatric ward by Isella et al. [19], which foundFig. 5 The sub-network corresponding to the PCA floors stratified by four occupational categories (red: administrative/support staff; blue: nurses;cyan: physicians; and green: oHCW)Table 6 Summary of number of visits per week and hours spent per week for each category in patient care (PCA), non-patient care(Non PCA), and mixed (mPCA) areasNon PCA mPCA PCAAdmin/Supp (N = 1031) Total number (and %) of floor visits per week 960 (30.2%) 1384 (43.6%) 831 (26.2%)Average time spent per floor (hrs/week) 6.3 8.0 8.4Other HCW (N = 452) Total number (and %) of floor visits per week 322 (15.9%) 791 (38.9%) 918 (45.2%)Average time spent per floor (hrs/week) 3.0 9.6 8.6Nurses (N = 1144) Total number (and %) of floor visits per week 522 (13.7%) 1577 (41.4%) 1709 (44.9%)Average time spent per floor (hrs/week) 1.2 7.1 16.9Physicians (N = 254) Total number (and %) of floor visits per week 130 (11.6%) 466 (41.6%) 525 (46.8%)Average time spent per floor (hrs/week) 1.4 8.0 11.1English et al. BMC Infectious Diseases  (2018) 18:184 Page 8 of 12physicians to have the least number of contacts of theoccupations surveyed and where a contact was definedas within 1.5 m for 20 s or more. In contrast, Polgreenet al. [17] found that nurses, resident physicians andfellows had the highest number of HCW contacts of thejob categories observed, where a contact was definedas within 0.9 m, but had no minimum time component(i.e., duration of contact). More recently, Mastrandrea[26], studying a single infectious disease ward usingradio frequency tracking devices, also found that physicianshad the highest number of contacts with other health careworkers, although this was within a total pool of only 22Table 7 HCW self-reported compliance with infection control guidelines and use of personal protective equipment (PPE)How regularly do you think colleagues comply with infection control guidelines (N = 2857) Mostly comply 1757 (61.5%)Partially comply 900 (31.5%)Poorly comply 200 (7%)On average, how regularly direct contacts with patients recorded in patient’s chart (N = 2897) Most often 1125 (38.8%)Sometimes 395 (13.6%)Not frequently 1226 (42.3%)N/A 151 (5.2%)How regularly record multiple contacts with same patient in patient’s chart (N = 2803) Most often 958 (34.2%)Sometimes 341 (12.2%)Not frequently 1300 (46.4%)N/A 204 (7.3%)Wear surgical/procedure mask when caring for patient with. .. (N = 3002) respiratory tract infection (RTI) 1414 (47.1%)TB N/AChickenpox N/AWear N95 respirator (not fit tested) when caring for patient with. .. RTI 171 (5.7%)TB 232 (7.7%)Chickenpox 75 (2.5%)Wear N95 respirator (fit tested) when caring for patient with. . . RTI 844 (28.1%)TB 1592 (53.0%)Chickenpox 345 (11.5%)Wear face or eye shield when caring for patient with. . . RTI 621 (20.7%)TB 1016 (33.8%)Chickenpox N/AWear one pair of gloves when caring for patient with. . . RTI 1757 (58.5%)TB N/AChickenpox 1618 (53.9%)Wear two pairs of gloves when caring for patient with. . . RTI 364 (12.1%)TB N/AChickenpox 273 (9.1%)Wear goggles when caring for patient with. . . RTI 352 (11.7%)TB 606 (20.2%)Chickenpox N/AWear gown when caring for patient with. . . RTI 1755 (58.5%)TB N/AChickenpox 1476 (49.2%)Wash hands when caring for patient with. . . RTI 2617 (87.2%)TB 2530 (84.3%)Chickenpox 2444 (81.4%)English et al. BMC Infectious Diseases  (2018) 18:184 Page 9 of 12HCWs. A study by Curtis et al. [16], used movementpatterns from electronic medical records to suggest thatresident physicians and nurses had the most frequentHCW contacts. While this conflicts with other findings,their definition of a contact differs significantly and did notinclude contacts in areas where electronic medical recordsfail to capture.Despite their lower HCW contact rate in our study,physicians may still play a key role in infection-relatedevents in the hospital. For example, significantly morephysicians reported direct patient contacts of > 20 patientsper day and were most likely to work in an additional butseparate healthcare facility. This indicates that physiciansmay have a higher capacity to facilitate disease spread thatpropagates across wards and from hospital to hospital.Nurses reported the most extended contact with patients,and so may be at a higher risk of becoming infected by apatient. On the other hand, due to their more localizedwork space (typically a single ward), they may have areduced role in the spreading of disease throughout thehospital population.Location analyses showed that public spaces, includingthe cafeteria, lobby café, and coffee shops, were visitedthe most frequently per week but for a relatively shorterduration of time; this finding highlights a potentialvulnerability of non-clinical spaces in healthcare facilitiesto promote infection spread for moderately- to highlytransmissible pathogens. Given the vast overlap of HCWs,patients and the general public that may simultaneouslyvisit these areas, disease spread could easily be facilitatedbetween otherwise unconnected wards or units (or thecommunity at large). Targeting these high-traffic areaswith interventions such as hand-hygiene (washing stationsor alcohol-based sanitizers), or mask distribution, or facili-tating spatial separation may be effective in reaching alarge and diverse subset of the hospital population.Inpatient locations were found to have a greater numberof visits per week compared to outpatient locations. Inpa-tients settings have patients with a higher acuity of illness,and therefore a greater diversity of HCWs may be in contactwith the patient. This suggests that there is an increasedrisk of disease spread in inpatient settings compared tooutpatient settings.Variable compliance in implementing and incorrectapplication of IPC precautions combined with the non-intuitively diverse structure of HCW contact patternsshown above, may lead to complex infection transmis-sion dynamics pathways. Accounting for this complexitywill require the use of quantitative complexity sciencetechniques that go beyond basic statistical description ofsurvey data.As with any paper-based questionnaire, one of thelimitations of this study was that the responses reliedon an individual’s recollection of movement throughoutthe hospital. To minimize the impact of this limitation,the respondents were given the choices of providingtheir contact history data based on a “typical week” ofwork, or “last week” of work, or “the last full weekworked”. After the paper questionnaires were distributedwithin participating hospitals, drop boxes were providedfor several weeks at study sites to collect responses. Weassume among those who selected “last week”, somemight have had a chance to assess their responses in “realtime”, while others relied on their immediate-past, or pastmemories (typical week).In the questionnaire, we clarified direct contacts asthose that occur “within 1 meter/3 feet”, while indirectcontacts are those that occur “within the same room butnot closer than 1 meter/3 feet”. Although based on thesedefinitions, these two types of contact are mutuallyexclusive, it might have been difficult for respondents tostrictly apply these definitions when recalling (immediate)past events. It is worth noting that in addition to theduration of contact, the type and intensity of contact areamong factors to be considered, as physical contact mightplay an important role for some HCAIs [22, 23]. In thispaper, our goal was not to construct direct contactnetworks between HCWs (i.e., all nodes in the networkrepresenting HCWs); rather, we presented co-locationnetworks (i.e., bipartite HCW-location networks) derivedfrom survey data. As such, we used each hospital floor asa single node in networks for ease of presentation. Toestablish formal inter-HCW contact networks, for out-break and transmission dynamics analysis, future studieswill utilize CONNECT I’s more refined data correspondingto smaller spatial units (please see Additional file 1) thanfloor-aggregated data.ConclusionThe network structures presented in this paper reveal ahigh degree of heterogeneity across HCW occupationsand their roles on different wards/floors. These intrica-cies combined with heterogeneity in implementing IPCmeasures imply that designing policy requires employingnetwork-based quantitative tools beyond that of basicaggregate statistics. These tools provide greater optionsand flexibility based on specific contact patterns thatfacilitate communicable disease transmission withinhospital settings. Future research based on heterogeneityof movement patterns across the 3 hospitals will allowfurther tailoring of interventions for setting-specific controlstrategies.The CONNECT I study provides insight into themovement and contact patterns of healthcare workers inthe general hospital setting. These results can informmodeling initiatives to more accurately simulate the spreadof HAI, and to optimize control strategies.English et al. BMC Infectious Diseases  (2018) 18:184 Page 10 of 12Additional filesAdditional file 1: CONNECT I Questionnaire. Description of data: Ageneric version of site-specific questionnaire used to collect data fromparticipating hospitals. (PDF 141 kb)AbbreviationsANOVA: Analysis of variance; CIHR: Canadian Institutes of Health Research;HAI: Healthcare-associated infection; HCW: Health care worker; IPC: Infectionprevention and control; MERS: Middle East respiratory syndrome;mPCA: Mixed patient-care area; Non-PCA: Non-patient-care area;oHCW: Other HCWs (non-physician, non-nurse); PCA: Patient-care area;RFID: Radio frequency identification; RSV: Respiratory syncytial virus;SARS: Severe acute respiratory syndromeFundingThis work was supported by grants from the Canadian Institutes of HealthResearch (CIHR). The funding agency had no role in the design of the study,or collection, analysis, and interpretation of data, or drafting the manuscript.Availability of data and materialsThe datasets generated and/or analysed during the current study are notpublicly available due to the inclusion of hospital administrative and humanresources data. Data are, however, available from the corresponding authorupon reasonable request and with permission of the study participating sites.Authors’ contributionsKME, BH and BP made substantial contributions to conception and design ofthe study; JML, AM, SAH and LJ made substantial contributions to thedeveloping the survey, implementation of the study in participating sitesand acquisition of data; KME, JML, AM, NH, RT, BH, SAH, LJ and BP madesubstantial contributions to the analysis and interpretation of data, and havebeen involved in drafting the manuscript. All 9 authors have seen andapproved this manuscript.Ethics approval and consent to participateThe University of British Columbia Behavioural Research Ethics Board, the IWKHealth Centre Research Ethics Board, the Mount Sinai Hospital ResearchEthics Board, and the Nova Scotia Health Authority Research Ethics Boardapproved this study. An information letter accompanied each questionnaire,detailing the survey protocol and objectives. The letter clarified that staff’svoluntary and anonymous participation, by returning the questionnaires indesignated drop boxes, implied their consent.Competing interestsThe authors declare that they have no competing interests.Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.Author details1Institute for Resources, Environment and Sustainability, University of BritishColumbia, 2202 Main Mall, Vancouver, BC V6T 1Z4, Canada. 2Departments ofPediatrics, and Community Health & Epidemiology, Canadian Center forVaccinology, IWK Health Centre, Nova Scotia Health Authority, DalhousieUniversity, Halifax, NS B3K 6R8, Canada. 3Mount Sinai Hospital, 600 UniversityAvenue, Toronto, ON M5G 1X5, Canada. 4Weill Cornell Medicine, 402 East 67St, New York, NY 10065, USA. 5Department of Pathology & LaboratoryMedicine, And Provincial Laboratory for Public Health of Alberta, 3030Hospital Drive NW, Calgary, AB T2N 4W4, Canada. 6British Columbia Ministryof Health, 1515 Blanshard St, Victoria, BC V8W 9P4, Canada. 7Departments ofPediatrics, and Microbiology & Immunology, Canadian Center forVaccinology, IWK Health Centre, Nova Scotia Health Authority, DalhousieUniversity, Halifax, NS B3K 6R8, Canada. 8Department of Medicine, DalhousieUniversity & Nova Scotia Health Authority, Halifax, NS B3H 1V7, Canada.Received: 21 August 2017 Accepted: 12 April 2018References1. Zoutman DE, Ford BD, Bryce E, Gourdeau M, Hébert G, Henderson E, et al.The state of infection surveillance and control in Canadian acute carehospitals. Am J Infect Control. 2003;31:266–73.2. Umscheid CA, Mitchell MD, Doshi JA, Agarwal R, Williams K, Brennan PJ.Estimating the proportion of healthcare-associated infections that arereasonably preventable and the related mortality and costs. Infect ControlHosp Epidemiol. 2015;32:101–14.3. Marschang S, Bernardo G. Prevention and control of healthcare-associatedinfection in Europe: a review of patients' perspectives and existingdifferences. J Hosp Infect. Elsevier Ltd. 2015;89:357–62.4. Public Health Agency of Canada. Routine practices and additionalprecautions for preventing the transmission of infection in healthcaresettings; 2012. p. 1–195.5. CDC. SPECIAL ARTICLE guideline for infection control in health carepersonnel, 1998; 1998. p. 1–66.6. Pourbohloul B, Meyers LA, Skowronski DM, Krajden M, Patrick DM, BrunhamRC. Modeling control strategies of respiratory pathogens. Emerg Infect Dis.2005;11:1249–56.7. Meyers LA, Pourbohloul B, Newman MEJ, Skowronski DM, Brunham RC.Network theory and SARS: predicting outbreak diversity. J Theor Biol. 2005;232:71–81.8. Mossong J, Hens N, Jit M, Beutels P, Auranen K, Mikolajczyk R, et al. In: RileyS, editor. Social contacts and mixing patterns relevant to the spread ofinfectious diseases, PLoS med, vol. 5; 2008. p. e74.9. Zagheni E, Billari FC, Manfredi P, Melegaro A, Mossong J, Edmunds WJ.Using time-use data to parameterize models for the spread of close-contactinfectious diseases. Am J Epidemiol. 2008;168:1082–90.10. Smieszek T, Barclay VC, Seeni I, Rainey JJ, Gao H, Uzicanin A, et al. Howshould social mixing be measured: comparing web-based survey andsensor-based methods. BMC Infect Dis. 2014;14:641.11. Smieszek T, Castell S, Barrat A, Cattuto C, White PJ, Krause G. Contact diaries versuswearable proximity sensors in measuring contact patterns at a conference:method comparison and participants’ attitudes. BMC Infect Dis. 2016. p. 1–14.https://doi.org/10.1186/s12879-016-1676-y.12. Rossana Mastrandrea JFAB. Contact patterns in a high school: a comparisonbetween data collected using wearable sensors, contact diaries andfriendship surveys; 2015. p. 1–26.13. Bernard H, Fischer R, Mikolajczyk RT, Kretzschmar M, Wildner M. Nurses’contacts and potential for infectious disease transmission. Emerg Infect Dis.2009;15:1438–44.14. Ueno T, Masuda N. Controlling nosocomial infection based on structure ofhospital social networks. J Theor Biol. Elsevier Ltd. 2008;254:655–66.15. Cusumano-Towner M, Li DY, Tuo S, Krishnan G, Maslove DM. A socialnetwork of hospital acquired infection built from electronic medical recorddata. J Am Med Inform Assoc. 2013;20:427–34.16. Curtis DE, Hlady CS, Kanade G, Pemmaraju SV, Polgreen PM, Segre AM. In:Colizza V, editor. Healthcare worker contact networks and the prevention ofhospital-acquired infections, PLoS ONE, vol. 8; 2013. p. e79906–11.17. Polgreen PM, Tassier TL, Pemmaraju SV, Segre AM. Prioritizing healthcareworker vaccinations on the basis of social network analysis. Infect ControlHosp Epidemiol. 2015;31:893–900.18. Hornbeck T, Naylor D, Segre AM, Thomas G, Herman T, Polgreen PM. Usingsensor networks to study the effect of peripatetic healthcare workers on thespread of hospital-associated infections. J Infect Dis. 2012;206:1549–57.19. Isella L, Romano M, Barrat A, Cattuto C, Colizza V, Van den Broeck W, et al.In: Cowling B, editor. Close encounters in a pediatric Ward: measuring face-to-face proximity and mixing patterns with wearable sensors, PLoS ONE, vol.6; 2011. p. e17144–10.20. Vanhems P, Barrat A, Cattuto C, Pinton J-F, Khanafer N, Régis C, et al. In: ViboudC, editor. Estimating potential infection transmission routes in hospital wardsusing wearable proximity sensors, PLoS ONE, vol. 8; 2013. p. e73970–9.21. Siegel JD, Rhinehart E, Jackson M, Chiarello L. 2007 guideline for isolationprecautions: preventing transmission of infectious agents in health caresettings. Am J Infect Control. 2007;35:S65–S164.22. Smieszek T. A mechanistic model of infection: why duration and intensity ofcontacts should be included in models of disease spread. Theor Biol MedModel. 2009;6:25–10.English et al. BMC Infectious Diseases  (2018) 18:184 Page 11 of 1223. De Cao E, Zagheni E, Manfredi P, Melegaro A. The relative importance offrequency of contacts and duration of exposure for the spread of directlytransmitted infections. Biostatistics. 2014;15:470–83.24. Haas CN, Rose JB. Gerba CP. Quantitative Microbial Risk Assessment: Wiley; 2014.25. Raboud J, Shigayeva A, McGeer A, Bontovics E, Chapman M, Gravel D, et al.In: Montgomery JM, editor. Risk factors for SARS transmission from patientsrequiring intubation: a multicentre investigation in Toronto, Canada, PLoSONE, vol. 5; 2010. p. e10717–0.26. Mastrandrea R, Soto-Aladro A, Brouqui P, Barrat A. Enhancing the evaluationof pathogen transmission risk in a hospital by merging hand-hygienecompliance and contact data: a proof-of-concept study. BMC ResearchNotes. BioMed Central. 2015. p. 1–11. https://doi.org/10.1186/s13104-015-1409-0.27. Raboud J, Saskin R, Simor A, Loeb M, Green K, Low DE, et al. Modelingtransmission of methicillin-resistant staphylococcus aureus among patientsadmitted to a hospital. Infect Control Hosp Epidemiol. 2016;26:607–15.English et al. BMC Infectious Diseases  (2018) 18:184 Page 12 of 12


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items