A Targeted Infection Prevention Intervention in Nursing Home Residents With Indwelling Devices: A Randomized Clinical Trial

Importance  Indwelling devices (eg, urinary catheters and feeding tubes) are often used in nursing homes (NHs). Inadequate care of residents with these devices contributes to high rates of multidrug-resistant organisms (MDROs) and device-related infections in NHs.

Objective  To test whether a multimodal targeted infection program (TIP) reduces the prevalence of MDROs and incident device-related infections.

Design, Setting, and Participants  Randomized clinical trial at 12 community-based NHs from May 2010 to April 2013. Participants were high-risk NH residents with urinary catheters, feeding tubes, or both.

Interventions  Multimodal, including preemptive barrier precautions, active surveillance for MDROs and infections, and NH staff education.

Main Outcomes and Measures  The primary outcome was the prevalence density rate of MDROs, defined as the total number of MDROs isolated per visit averaged over the duration of a resident’s participation. Secondary outcomes included new MDRO acquisitions and new clinically defined device-associated infections. Data were analyzed using a mixed-effects multilevel Poisson regression model (primary outcome) and a Cox proportional hazards model (secondary outcome), adjusting for facility-level clustering and resident-level variables.

Results  In total, 418 NH residents with indwelling devices were enrolled, with 34 174 device-days and 6557 anatomic sites sampled. Intervention NHs had a decrease in the overall MDRO prevalence density (rate ratio, 0.77; 95% CI, 0.62-0.94). The rate of new methicillin-resistant Staphylococcus aureus acquisitions was lower in the intervention group than in the control group (rate ratio, 0.78; 95% CI, 0.64-0.96). Hazard ratios for the first and all (including recurrent) clinically defined catheter-associated urinary tract infections were 0.54 (95% CI, 0.30-0.97) and 0.69 (95% CI, 0.49-0.99), respectively, in the intervention group and the control group. There were no reductions in new vancomycin-resistant enterococci or resistant gram-negative bacilli acquisitions or in new feeding tube–associated pneumonias or skin and soft-tissue infections.

Conclusions and Relevance  Our multimodal TIP intervention reduced the overall MDRO prevalence density, new methicillin-resistant S aureus acquisitions, and clinically defined catheter-associated urinary tract infection rates in high-risk NH residents with indwelling devices. Further studies are needed to evaluate the cost-effectiveness of this approach as well as its effects on the reduction of MDRO transmission to other residents, on the environment, and on referring hospitals.

Trial Registration  clinicaltrials.gov Identifier: NCT01062841

Approximately 1.4 million Americans reside in 15 600 US nursing homes (NHs), and in 2010 there were 2.5 million short-stay admissions.¹ More than 40% of all Medicare beneficiaries discharged from hospitals in 2011 required postacute care.² Multidrug-resistant organisms (MDROs) are endemic in NHs, with prevalence rates exceeding 35% and surpassing those for hospitals.^3-7 With increasing acuity of illness, the risk of acquiring new infections increases substantially, with approximately 2 million infections occurring in NHs each year.⁸

Indwelling devices, such as urinary catheters and feeding tubes, are frequently used in NHs.^9-13 National data for the United States show that approximately 5% to 7% of all NH residents have an indwelling urinary catheter and that 5% to 7% have feeding tubes, while 12% to 15% of new admissions to NHs have an indwelling urinary catheter.^1,9-12 Compared with NH residents without indwelling devices, residents with indwelling devices have a higher prevalence of MDRO colonization at multiple anatomic sites.^3,14-16 The NH residents with indwelling devices share many characteristics with hospitalized populations, and improper care of residents with these devices presents opportunities for pathogen acquisition. The hands of health care workers (HCWs) in these NHs are frequently colonized with gram-negative bacilli (66%), Candida (41%), Staphylococcus aureus (20%), and vancomycin-resistant enterococci (VRE) (9%), increasing the risk of pathogen transmission when providing assistance with various activities of daily living.¹⁷

Despite the large number of NH residents who are colonized and subsequently infected, randomized trials evaluating interventions focused on the entire population or specific high-risk groups are infrequent.^18,19 Studies conducted at single NHs have generally focused on 1 of the following 4 types of interventions: hand hygiene,^17,20,21 gown use and contact precautions,²² decolonization regimens,²³ or infection prevention education.^24,25 Although these studies show a trend toward reduced MDRO colonization and infections, the reduction in MDROs has not been statistically significant in most cases. Therefore, we conducted a multicenter randomized clinical trial to test whether a multimodal practical evidence-based targeted infection program (TIP) would reduce the prevalence density of MDRO colonization and the incidence of new device-related infections in high-risk NH residents.

We approached 17 community-based NHs in southeast Michigan that had expressed a preliminary interest in participating in our research program (Figure). Thirteen agreed to have initial conversations with us, of which 12 agreed to participate and remained in our study for 3 years from May 2010 to April 2013. This study was approved by the institutional review board at the University of Michigan. The full study protocol can be found in the trial protocol in Supplement 1. All Medicare-certified and Medicaid-certified NHs have an infection control program.²⁶ Similar to most US NHs, all study sites had an on-site infection preventionist, most of whom served in this role part-time and had additional responsibilities (eTable 1 in Supplement 2). All NHs had access to laboratory and radiology services.²⁷ The TIP intervention included the following: (1) preemptive barrier precautions; (2) active surveillance for MDROs and infections, with data feedback; and (3) NH staff education on key infection prevention practices and hand hygiene promotion (Table 1). The control group NHs continued to practice according to their own infection prevention policies.

Written informed consent to collect surveillance samples and resident-specific data was obtained from each resident or his or her durable power of attorney. Inclusion criteria were (1) any short-stay or long-stay resident with an indwelling urinary catheter, a feeding tube (nasogastric or percutaneous endoscopic gastrostomy tube), or both for more than 72 hours and (2) informed consent. Residents receiving end-of-life care were excluded. Study personnel (B.L., S.E.M., J.F., and E.K.) obtained data on participant characteristics, age, sex, functional status,²⁸ comorbidity,²⁹ and facility-level predictors, such as the quality ratings.³⁰ The Centers for Medicare & Medicaid Services created the 5-star quality rating system to help consumers, their families, and caregivers compare NHs. This rating incorporates health inspections, staffing, and quality measures. While the field staff (B.L., S.E.M., J.F., and E.K.) were aware of the intervention assignment, one of us (K.S.) processing microbiology cultures was blinded.

Our sample included 12 NHs, with a mean of 137 beds each. The enrollment period for the study was 36 months. Based on preliminary data,^9-11 we estimated that each NH would care for a mean of 22 residents with an indwelling device per year. Power calculations were based on anticipating a 30% reduction (rate ratio, 0.70) in MDRO prevalence as a result of our intervention. The desired statistical power was 80%, and the significance criterion was .05. Based on previous data,³¹ the expected MDRO prevalence density rates were 42 MDRO isolates, including new and repeat isolates from all anatomic sites, per 1000 device-days in the intervention group and 60 MDROs per 1000 device-days in the control group. Because this was a prospective study of NH residents, we also adjusted for intracluster correlation (using an intraclass correlation coefficient of 0.07).^32-38 Therefore, 12 NHs were sufficient to reach 80% power to reject the null hypothesis of no difference in the overall MDRO prevalence density between the 2 groups. We anticipated conducting 1600 study visits over 3 years.

We chose to randomize NHs instead of participants using cluster randomization. Randomizing participants can lead to contamination by HCW experiences learned when caring for the intervention participants, thus underestimating the true effect of the intervention. Therefore, NHs served as the unit of allocation. The NHs were stratified by ownership (for profit vs not for profit) because prior studies^39,40 have shown differences in quality of care by ownership status. Once stratified, NHs were assigned to intervention status using computer-generated randomization by our study statistician (A.G.). Six were randomized to the intervention group and 6 to the control group.

Individuals in the intervention group with an indwelling device, irrespective of their MDRO colonization or infection status, were placed on preemptive barrier precautions for the duration of their participation. Specifically, barrier precaution signs were placed on the doors to their rooms, inside their closet, at the nurses’ station, and on their medical records. The HCWs were encouraged to perform hand hygiene before and after providing any care to the participants and were encouraged to wear gowns and gloves when performing morning and evening care, device care, and during any care activity where splashing may be expected to occur. Hand hygiene products, disposable gowns, and personal protective equipment caddies were placed strategically at each intervention site. Residents were not isolated. They were allowed to socialize and obtain rehabilitation outside their room. For nonenrolled residents, intervention NHs followed their existing policies for those infected or colonized with MDROs.

Microbiological samples from the nares, oropharynx, enteral feeding tube insertion site, suprapubic catheter site, groin, perianal area, and wounds were obtained from all participants to assess MDRO colonization at baseline, day 15, and then monthly for up to 1 year, regardless of prior colonization status. Because there is no established active surveillance strategy in this diverse population with varying lengths of stay and case mix and because residents with devices have multisite colonization, we cultured multiple anatomic sites to assess our primary outcome.^41-43 Standard microbiological methods were used to identify methicillin-resistant S aureus (MRSA), VRE, ceftazidime-resistant gram-negative bacilli, and ciprofloxacin-resistant gram-negative bacilli by trained research staff as previously described.³¹ Infections were identified using clinical definitions from participants’ medical records, abstracted by trained study coordinators (B.L., S.E.M., and J.F.). The MDRO colonization and infection rates were reported monthly to the intervention NHs by the study coordinators. Graphs, charts, and tables of the data were designed to be easily understood by HCWs and emphasized infection and MDRO risks posed by indwelling devices.

Once a resident was enrolled in the study, the nurse’s aide, nurse, his or her physician of record, and the infection preventionist received informational pocket cards. The cards focused on infection recognition using NH-appropriate definitions, including the minimum criteria to initiate antibiotics for specific infections (eg, catheter-associated urinary tract infections [CAUTI])⁴⁴ as well as criteria by McGeer et al⁴⁵ for surveillance of infections (eAppendix in Supplement 2). Infection preventionists at the intervention NHs were also invited to a half-day conference on surveillance methods for infections (Conference Agenda in eAppendix in Supplement 2). In addition, a structured, interactive infection control education program for HCWs was designed, implemented, and evaluated. Ten educational modules, executed every 2 to 3 months over 3 years, were conducted by the research team (L.M., B.L., S.E.M., E.K., R.A.R., R.N.O., and S.F.B.) at the intervention NHs. The educational sessions were targeted to nurses and nurses’ aides, although all NH staff were encouraged to attend. Each session was 30 minutes long, was offered multiple times to include workers on all shifts, and was accompanied by a pretest and posttest consisting of 5 to 10 questions. Session topics included an overview of infection prevention practices, hand hygiene, appropriate indications for device use and device care, and recognition of infections, with content following evidence-based guidelines. Each didactic session was presented using pretaped DVDs, followed by an interactive component that included game formats modeled after Jeopardy and Wheel of Fortune or device care skits on infection prevention themes. The HCWs at the control facilities were given a similar test on each session at the same time, but no education was provided.

Barrier precautions were promoted using posters, videos, and dance routines. The HCWs were educated on the indications, technique, and duration of effective hand hygiene as well as appropriate glove and gown use. To reduce the Hawthorne effect (also referred to as the observer effect), structured 30-minute observations to monitor HCW activities and their use of barrier precautions were conducted within participant rooms. The alternative of following an HCW in this setting would have changed his or her behavior.

The primary study outcome was the overall MDRO prevalence density rate, defined as each participant’s total number of MDRO-positive anatomic sites across all MDROs per visit averaged over the duration of his or her participation. While incidence rates describing new acquisitions are appropriate for acute care hospitals, prevalence measures are better suited to describe the prevalence of multianatomic site MDRO colonization in a long-stay NH population from a public health perspective.^42-44,46 This measure captures both transient and persistent MDRO colonization over the duration of participation (ie, a participant with persistent MDRO colonization would have a higher prevalence than someone with intermittent or no colonization). In addition, participants with a greater involvement of MDRO across anatomic sites would have a higher prevalence. We also evaluated the effect of this intervention on an individual’s risk of new MDRO acquisition, defined as the number of residents with new acquisitions per 1000 device-days at risk as exploratory analyses. New acquisition was assessed at the resident level (eg, a resident culture negative for MRSA at baseline with a subsequent follow-up culture positive for MRSA at any anatomic site). The secondary outcome was the incidence of device-associated infections per 1000 device-days. We defined clinical infections objectively by the presence of both of the following: (1) a clinical note in the participant’s medical record documenting an infection and (2) prescription of a systemic antibiotic for at least 3 days to treat that infection. Each participant was followed from enrollment up to 1 year or through his or her discharge, death, or device removal, whichever occurred first.

A mixed-effects multilevel Poisson regression model was used to predict MDRO prevalence density as a function of the intervention (primary outcome), including facility as a random effect and offset by the number of anatomic sites sampled. To obtain the geometric mean, the prevalence measures were log transformed and then exponentiated back to the original units. Because we expected participants colonized at baseline to have a greater likelihood of subsequent colonization, we included participant-level baseline MDRO colonization as a fixed effect. The model was adjusted for participant-specific covariates (participant-level baseline colonization with the specific MDRO, age, sex, race, and length of stay before enrollment), and NH quality ratings.³⁰ Using a Cox proportional hazards model, we conducted exploratory analyses to evaluate the effect of this intervention on an individual’s risk of new MDRO acquisition, defined as the number of residents with new acquisitions per 1000 device-days at risk after adjusting for resident-level and facility-level covariates, as well as clustering by facility. Residents colonized with the specific MDRO at baseline were excluded from these analyses. For the secondary outcome of incident device-associated infections, a Cox proportional hazards model using the cluster option with robust standard errors was used to predict time to the first and recurrent infections, adjusting for participant-level and facility-level covariates. Residents with only a baseline visit and no follow-up visits were excluded from analysis.

In total, 203 NH residents with indwelling devices were enrolled in the intervention group, with 17 490 device-days and 3283 active surveillance samples collected. In total, 215 NH residents were enrolled in the control group, with 16 684 device-days and 3274 samples collected. The exclusion of residents with baseline visits only and no follow-up resulted in an analytic population of 154 residents with 871 study visits in the intervention group (with 3116 samples collected) and 162 residents with 857 visits in the control group (with 3072 samples collected). There were no significant differences in baseline characteristics between these 2 groups (Table 2). The NH characteristics, assignments, and resident follow-up for both groups per Consolidated Standards of Reporting Trials guidelines⁴⁷ are summarized in the Figure and in eTable 1 and eTable 2 in Supplement 2, respectively. The number of surveillance samples collected during the study and from specific anatomic sites did not differ between the 2 groups (eTable 3 and eTable 4 in Supplement 2).

The MDRO prevalence density rates based on cluster-level summaries are listed in eTable 5 in Supplement 2. Using data from the participants enrolled during the first 30 days, there was no statistically significant difference in the overall baseline MDRO colonization rates between the groups.

For the primary outcome, when adjusted for the cluster study design as well as for baseline colonization, age, sex, race, and NH quality rating, we found a 23% reduction in the prevalence density rate for all MDROs (rate ratio, 0.77; 95% CI, 0.62-0.94; P = .01) in the intervention group (Table 3). When analyzing data for individual MDROs, the intervention group had a 22% reduction in the MRSA prevalence density rate (rate ratio, 0.78; 95% CI, 0.64-0.96; P = .01). Participants with urinary catheters showed reduced colonization with MRSA and ceftazidime-resistant gram-negative bacilli, and participants with feeding tubes showed reduced colonization with MRSA. We also assessed the effectiveness of the TIP intervention in reducing MDRO colonization at different anatomic sites (eTable 6 in Supplement 2). When evaluating anatomic site–level differences for individual MDROs, the intervention group showed a lower mean MDRO prevalence (eTable 7 in Supplement 2)

New acquisitions of MRSA and resistant gram-negative bacilli were common. For example, after excluding residents with baseline MRSA colonization, the rates of new acquisition were 6.2 and 7.9 per 1000 device-days at risk at intervention and control sites, respectively (Table 4). New VRE acquisition occurred less frequently. After excluding residents with baseline VRE colonization, the rates of new acquisition were 1.7 and 2.3 per 1000 device-days at risk at intervention and control sites, respectively. After accounting for facility-level and resident-level covariates as well as facility-level clustering, the hazard ratio for new MRSA acquisition was lower for the intervention group than for the control group (hazard ratio, 0.78; 95% CI, 0.65-0.95; P = .01). There were no differences in new VRE or resistant gram-negative bacilli acquisitions.

As summarized in Table 5, there were 56 new episodes of clinically defined CAUTIs in the intervention group over 9413 catheter-days, with an incidence rate of 5.9 per 1000 catheter-days. There were 75 new CAUTIs in the control group over 8118 catheter-days, with an incidence rate of 9.2 per 1000 catheter-days (cluster- and covariate-adjusted hazard ratio, 0.69; 95% CI, 0.49-0.99; P = .045). In total, 74 first new CAUTIs were observed, including 31 in the intervention group and 43 in the control group (crude incidence of CAUTIs, 5.2 and 10.0 per 1000 catheter-days, respectively). Based on the results of the Cox proportional hazards multivariable regression analyses, which focused on the first new clinically defined CAUTIs, the risk of having a CAUTI remained significantly lower in the intervention group than in the control group (hazard ratio, 0.54; 95% CI, 0.30-0.97; P = .04). There were much fewer feeding tube–associated pneumonias and skin and soft-tissue infections, with no differences between the intervention and control groups.

Our implementation process was evaluated by assessing the use of NH-appropriate definitions to monitor infection rates, post–in-service HCW knowledge scores, and use of barrier precautions during high-risk activities, such as morning and evening care or when splashing was expected. These process measures are listed in eTable 8 in Supplement 2.

We found that a multimodal evidence-based TIP intervention involving preemptive barrier precautions, active surveillance for infections and MDROs, hand hygiene promotion, and structured infection prevention education reduced colonization with MDROs and prevented clinically defined CAUTIs in our NH population with indwelling devices. We demonstrated the utility of using a comprehensive yet technically simple strategy to produce these outcomes instead of time-consuming and resource-intensive measures. Our strategies were aimed at enhancing the existing infection prevention program in NHs by making education interactive, multidimensional, and directly applicable to a high-risk population.

Several factors help explain the observed reduction in MDROs. First, preemptive barrier precautions likely led to reduced pathogen prevalence density and new MRSA acquisition. Instituting preemptive precautions has led to reduced MRSA, Clostridium difficile, and resistant gram-negative bacilli colonization in acute care, usually during MDRO outbreaks.⁴⁸ Second, we complemented preemptive barrier precautions with interactive, engaging infection prevention education delivered through HCW in-service implementation. Informational slides were followed by creative interactive games that reinforced educational messages presented during the didactic sessions. Third, our intervention included interactive methods of enhancing hand hygiene, such as dance routines; demonstrations of hand washing technique (using Glo Germ gel; Glo Germ Company); and pre-post hand cultures, as well as environmental sampling to illustrate the transfer of organisms. These interactive strategies enhanced HCW knowledge and participation, facilitated conversations about infection prevention in NHs, and likely led to reduced transmission of pathogens when providing indwelling device care.

Our multimodal TIP intervention also led to a reduction in clinically defined CAUTIs, reflecting a reduction in antibiotic use as the intervention NHs adopted appropriate criteria for diagnosis of infections during the study period. Distribution of pocket cards that focused on infection recognition to all health care professionals involved in the resident care, including nurses’ aides, nurses, and physicians, may have also affected rates by improving diagnosis of CAUTIs and reducing antibiotic use, leading to decreased prevalence density rates of MDROs. Educational sessions on appropriate urinary catheter use and care (including emphasis on hand hygiene) may have reduced CAUTI rates by limiting contamination of catheter sites. In the Netherlands, Cools and van der Meer⁴⁹ established an overall infection control program in one NH, with emphasis on hand hygiene, restriction of long-term indwelling urethral catheters, treatment of only symptomatic infections, and weekly all-physician meetings to discuss residents with infections. That program reduced clinically defined urinary tract infections from 256 in 515 residents in year 1 to 66 in 527 residents in year 6. In a more recent study, Evans et al²⁵ introduced an MRSA bundle of universal nasal surveillance on admission, transfer, or discharge, as well as contact precautions for MRSA infection or colonization, hand hygiene promotion, and institutional culture change. Their program resulted in a trend toward reducing MRSA CAUTIs in Veterans Affairs spinal cord injury units. Ours is one of the first studies involving multiple community-based NHs in the United States to show the effectiveness of a multimodal TIP intervention in reducing antibiotic-treated CAUTIs.

The key strengths of this study are the evaluation of a multimodal TIP intervention in NHs and the goal to reduce MDRO colonization and infections in a high-risk population with indwelling devices. The TIP components were designed to upgrade and integrate individual practices into routine NH care, with emphasis on appropriate use and harms related to the use of indwelling devices. We examined a high-risk population in which there is a lack of rigorous randomized clinical trials to guide clinical practice and policy.¹⁷ Ours is also one of the initial studies involving a community-based NH consortium showing that horizontal interventions to enhance routine infection prevention practices reduce MDRO colonization and antibiotic use related to CAUTIs in a high-risk population.

The study also has limitations. First, our study was conducted in NH facilities in southeast Michigan. While our study facilities resemble other US NHs with respect to their ownership status, the mean number of beds, availability of laboratory and radiology services, and the presence of a person responsible for infection prevention, our results may not be generalizable to other types of long-term care facilities, such as assisted-living facilities and hospice centers. Second, we did not assess the effect of the TIP intervention on other potentially at-risk residents at these NHs (eg, residents with wounds). With the increasing complexity of resident populations being served at these NHs, future studies should target other at-risk groups. Our goal was to evaluate the effectiveness of our TIP intervention in reducing MDRO transmission from HCWs to a population at high risk of MDRO acquisition due to the presence of indwelling devices. Whether these interventions lead to reduced transmission from these high-risk residents to the rest of the population also needs to be evaluated in the future. Third, we defined CAUTIs using clinical definitions. A recent systematic review demonstrates that there are many different definitions for CAUTIs used in the long-term care setting.⁵⁰ Ninety percent of all controlled interventional studies reported either symptomatic or physician-diagnosed urinary tract infections (n = 13) or CAUTI definition not specified (n = 5) as the outcome of interest. We chose to use clinically defined CAUTIs because this measure represents actual antibiotic-treated infections, and antibiotic use remains a significant risk factor for MDRO colonization. Fourth, we did not monitor hand hygiene that could have occurred outside a resident’s room; therefore, our estimates of in-room hand hygiene compliance may be conservative. To minimize the Hawthorne effect, research personnel conducted in-room observations instead of following HCWs as they conducted their duties.⁵¹ Fifth, we did not assess the cost-effectiveness of the interventions because it was beyond the scope of this study.

Limitations notwithstanding, we found that a comprehensive multimodal TIP intervention in NHs focused on reducing MDRO colonization and CAUTIs is feasible and effective. We designed the TIP intervention by engaging community-based NHs with little experience in conducting prospective research and with few infection prevention resources to demonstrate strategies that could be adopted in this traditionally resource-poor setting.

Accepted for Publication: December 14, 2014.

Corresponding Author: Lona Mody, MD, Division of Geriatric and Palliative Medicine, Department of Internal Medicine, University of Michigan Medical School, 300 N Ingalls Rd, Room 905, Ann Arbor, MI 48109 (lonamody@umich.edu).

Published Online: March 16, 2015. doi:10.1001/jamainternmed.2015.132.

Author Contributions: Dr Mody had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Mody, Krein, Saint, Galecki, Kauffman, Bradley.

Acquisition, analysis, or interpretation of data: Mody, Min, Lansing, McNamara, Symons, Fisch, Koo, Galecki, Kabeto, Fitzgerald.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Mody, Min, Galecki, Kabeto.

Obtained funding: Mody, Krein, Saint, Galecki, Kauffman, Bradley.

Administrative, technical, or material support: Mody, Lansing, McNamara, Fisch, Rye, Fitzgerald, Olmsted.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported by RO1AG032298 from the National Institute on Aging (Drs Mody, Krein, Saint, Galecki, Kauffman, and Bradley), by K23AG028943 from the National Institutes of Health (Dr Mody), by RO1AG041780 from the National Institute on Aging (Drs Mody, Min, Bradley), and by the Claude D. Pepper Older American Independence Centers funding from the National Institute on Aging (Dr Mody).

Role of the Funder/Sponsor: The sponsors of the study had no role in the design, data gathering, analysis, interpretation, writing of the report, or decision to submit the article for publication.

Additional Contributions: With approval of the National Institute on Aging, GOJO Industries, Inc donated personal-use hand sanitizer bottles to the intervention sites.

Correction: This article was corrected on April 30, 2015, to fix an error in the byline.