Study setting and population

Our study population comprised of medical doctors and nurses working in an ICU in the year 2015. The medical doctor and nurse could work in the same ICU and more than one reply was possible from one unit.

An ICU was defined as a unit meeting all of the following criteria: provides facilities for invasive mechanical ventilation, and pump-controlled administration of infusion, functions 24 h a day and 7 days a week, and there is at least one doctor immediately available at all times to deal with emergencies.

Questionnaire

An online questionnaire was developed and consisted of five parts: 1) Characteristics of the respondent and ICU setting, 2) Clinical practices for CL insertion, 3) Clinical practices for CL maintenance, 4) Monitoring of outcomes and processes and 5) Attitudes towards measurement as a tool for improvement. We used as a reference clinical practice guidelines published by the Society for Healthcare Epidemiology of America (SHEA) [4]. We included questions on the measurement of outcomes (CLABSI rate and the ability to report selected indicators) and measurement of processes such as compliance with prevention practices, including hand hygiene, and the device utilization ratio (ratio of central-line days to patient-days). Attitudes regarding the implementation of a data collection system was measured using a 5 point-Likert scale (1: strongly agree, 5: strongly disagree) [17, 18].

The questionnaire was developed and pretested in English and translated into 9 languages (Spanish, German, Portuguese, Italian, French, Dutch, Russian, Mandarin and Japanese). Native speaker intensive care doctors and/or infection control practitioners translated the questionnaire. Each translation was independently verified by a second native speaking physician. Participation was anonymous.

We used Limesurvey ® 2.0, an open source web survey application, to collect the data [19].

Dissemination to target group

The questionnaire was available online from 10 June 2015 to 31 October 2015. It was endorsed by 5 international societies (the European Society of Intensive Care Medicine (ESICM), the Society of Critical Care Medicine (SCCM), the World Federation of Societies of Intensive Care and Critical Medicine (WFSICCM), the International Symposium on Intensive Care and Emergency Medicine (ISICEM) and the Middle East Critical Care Assembly (MCCA)), and one national society (Japanese society of intensive care medicine (JSCIM)). Endorsement implied mailing to members; and/or posting on the website. We also identified and contacted national ICU societies who advertised the questionnaire on their website and pertaining congresses, and developed mass mailings to all its members.

Sample size

No sample size or power calculations were conducted.

Data analysis

Descriptive statistics were used to summarize characteristics of the study population Standard errors were calculated by dividing the standard deviations of each estimate by the square root of the sample size (n).

Based on the 2015 World Bank classification [20] we categorized countries as low, middle and high- income. We computed weighted estimates for middle and high-income countries using total country population [20] as the weight (to correct for contributing responses from each country) for those that provided at least 10 completed replies (arbitrary cut-off).

Analyses were conducted using STATA 13 (svyset and svy commands for survey data for weighted estimates).

Detailed country specific data are available as a Additional file 1. Data are freely available and have been deposited in the Dryad Digital Repository: http://dx.doi.org/10.5061/dryad.f7h12

Further use and exploitation of these data is encouraged.

Key results

This study represents, to our knowledge, the first international survey assessing CLABSI prevention practices (clinical and measurement), and attitudes towards measurement reported by ICU doctors and nurses. More than 3,000 provided complete responses, 40% of them were nurses.

A majority of respondents (80%) report the existence of CLABSI prevention guidelines in their ICU, demonstrating a wide awareness of the CLABSI problem. However several areas for improvement have been identified. Combined compliance to 4 recommended practices without the use of the 2 not-recommended practices (such as antimicrobial prophylaxis) at CL insertion was low particularly in middle income countries, the weakest point being in the utilization of sterile drape to cover patients from head-to-toe and the use of >0.5% chlorhexidine for skin preparation. Only a minority of respondents still report the not recommended femoral vein as the most used insertion site in their ICU. During maintenance, dressings are changed more often than recommended and assessment of need of the CL is not always done on a daily basis. Despite most of the respondents agreeing that measurement is essential for improvement, and routinely counting CLABSI, only few were able to report the actual CLABSI rate in their unit. Our findings are comparable to several other studies or surveys conducted on CLABSI prevention practices in ICU’s at local level [21–23]. A survey assessing implementation of preventive CLABSI measures and monitoring of compliance to these measures, identified that the use of full body drape during CVC insertion was among the measures less applied. Moreover, less than 50% of hospitals included in this study reported monitoring compliance to recommended measures [22]. Finally, Furuya et al. identified that daily line checks and optimal site selection were least commonly implemented. In this study only a disappointing 38% of those that monitored bundle implementation reported full compliance, this aligns with our low reported numbers. The availability of a bundle policy and monitoring of compliance and a compliance of at least 95% were needed to decrease CLABSI rates [23].

Strengths and limitations

The strengths of this study are the large number of respondents, the proportion of nurses (40%) and the ability to provide country-specific results for a large number of countries. However, our results are based on a non-random sample of respondents, we have not been able to compute a response rate, and we do not know the number of ICUs as the online questionnaire did not include questions allowing for the identification of the ICU, in order to preserve the anonymity of the respondents. In addition, given the dissemination strategy-through international and national societies-some categories of ICU doctors and nurses are likely to be overrepresented. Members of these societies might be better informed, and might have better prevention practices, than non-respondents to the survey. Those working in ICUs where CLABSI prevention guidelines exist might have been more motivated to participate in this survey. Reported practices are likely to differ from actual practices and be biased towards what the respondent believed to be desirable. These selection and reporting biases would lead to our results overestimating true compliance with recommended practices.

We used SHEA guidelines, but these are not “global” standards and some prevention measures recommended in SHEA guidelines might be controversial such as disinfection with chlorhexidine 0.5% in specific settings [24]. However it was beyond the scope of this article to assess the evidence supporting each recommendation. Detailed information for each recommendation is given in the country-specific file, so that the usefulness of this information can be assessed against national or local guidelines.

The grouping of our data in order to provide estimates for “high income” and “middle income” countries was a convenient way to aggregate the data, but given the heterogeneity of practices between countries these summary estimates have limited meaning, and the most relevant data can be found in the country-specific results presented in the Additional file 1.

We did not explore whether results differed between nurses and doctors (this will be the subject of a subsequent research).

Interpretation

The large number of respondents, and the large proportion of ICUs where CLABSI prevention guidelines are available, likely reflect the interest and awareness of the ICU community in the issue of CLABSI prevention. Selection and reporting bias in our study lead to the overestimation of CLABSI prevention practices in ICU’s, therefore weaknesses identified seem robust enough to identify interventions for improvement. Priorities for improvement differ between countries; from encouraging the use of chlorhexidine >0.5% in alcohol for skin preparation and sterile drapes to covering patients from head to toes, to discouraging the use of the femoral vein as preferred insertion site. Improving knowledge of clinical guidelines is far from sufficient to improve practices but it is a prerequisite [25].

Availability and discussion of ICU data is also very important as it can serve to highlight opportunities and areas for improvement. The majority of respondents agree that monitoring of CLABSI-related measures stimulates quality improvement but very few actually knew their data. In order for clinical staff to monitor trends over time and report real time feed-back they should be educated in how to generate reliable data through indicators that measure compliance. A compromise needs to be found between time-consuming data collection, and usefulness of data. Simple measures promoted by hospital management can be implemented to increase knowledge of CLABSI data at ICU level, e.g. a panel with the number of days since last ICU-acquired infection (including CLABSI) can be displayed on the wall and updated daily.