Point-prevalence surveys of antibiotic use and HAIs

The discovery of penicillin in 1928 by Alexander Fleming changed the treatment of infectious diseases. Since then, antibiotics have become one of the most important advances in medical treatment.¹ Antibiotics are among the most commonly prescribed drugs in both hospitals and the community.² There have been significant reductions in morbidity and mortality associated with the use of antibiotics since they were first introduced into medical practice.³ Although initially very effective, bacteria started to develop resistance to antibiotics rapidly after being first used by the public, and indeed antibiotic resistance has become increasingly prevalent in hospital and community healthcare settings across the world.⁴

Unfortunately, the ongoing discovery of new antibiotics has not continued and rapidly emerging antibiotic resistance threatens society as a whole.⁵ The resistance level in the countries of the European Union, particularly those in Southern Europe, is reaching alarming levels; such a worrying trend is also being noticed in Central Europe.^6,7 The risk of dissemination of antibiotic-resistant pathogens is growing; thus, the appropriate use of antibiotics is of increasing importance.^1,6 The inappropriate use of antibiotics and insufficient infection prevention and control practices have been shown to be a major cause for the development of several antibiotic-resistant bacteria.^8,9  In addition to antibiotic resistance, inappropriate antibiotic use may increase morbidity and mortality, risks of side effects and increased hospital costs.^10,11

Optimal prescribing of antibiotics is the first step in reducing antibiotic resistance.  Inappropriate use of antibiotics is associated with drug-resistant hospital-acquired infections (HAIs).^12,13 The role of antibiotic exposure in the development of HAI has been shown in many studies.^14–16 Moreover, HAI is a major contributor to prolonged hospital stay, increased healthcare cost, mortality and morbidity.^17–19 The level of HAIs within a particular institution is considered as an important indicator of the quality and safety of care.²⁰ HAIs vary from one hospital to another depending on the patient population and procedures used.¹³ It has been shown that HAI risk is associated with multiple factors (including patient’s age, length of hospital stay, an immune-compromised state, pre-existing underlying disease and the presence of invasive medical devices, such as urinary catheters, central lines, and ventilation) and that HAIs can be prevented through suitable preventative methods.^21,22

Identification of risk factors for HAI is important for many reasons: patients with known risks need special care, knowledge of the risk might aid early diagnosis and treatment in patients incubating an infection and observation data may be stratified according to risk factors enabling internal and external comparison of infection rates.²³  Point-prevalence surveys provide useful data on antibiotic use and HAI.² Surveillance systems that detail antibiotic use and HAI prevalence are important approaches to identify areas for intervention and to target quality improvement interventions to deal with the problem of antibacterial resistance.²⁴

A total of five acute teaching hospitals participated in the two point-prevalence studies. These hospitals represent five of the main acute care hospitals in the five Health and Social Care (HSC) Trusts present in Northern Ireland at the time the research was conducted. These five HSC Trusts served a total population of approximately 1.7 million, as follows: Antrim Area Hospital (426 beds) in the Northern Health and Social Care Trust (NHSCT); Ulster Hospital (561 beds) in the South Eastern Health and Social Care Trust (SEHSCT); Altnagelvin Hospital (495 beds) in the Western Health and Social Care Trust (WHSCT); Belfast City Hospital (900 beds) in Belfast Health and Social Care Trust (BHSCT); and Craigavon Hospital (426 beds) in Southern Health and Social Trust (SHSCT).

These hospitals all provide acute, general medical and surgical services, support a range of outpatient facilities and act as centres for the coordination of health service provision throughout their defined geographical areas in Northern Ireland.

The participating hospitals during the first point-prevalence study (June 2011) were: Antrim Area Hospital, Ulster Hospital, Altnagelvin Hospital and Belfast City Hospital, while the participating hospitals during the second point-prevalence study (January 2012) were: Antrim Area Hospital, Ulster Hospital, Altnagelvin Hospital and Craigavon Hospital. A total of six wards in each hospital were selected to further investigate the risk factors of HAI; these wards were the adult intensive care unit, one general medical ward, one general surgical ward, a respiratory ward, a care of the elderly ward and a paediatric ward. Craigavon Hospital did not participate in investigating the risk factors of HAIs.

The prospective point-prevalence surveys took place on two different dates in the summer (June 2011) and in the winter (January 2012). Data were collected in each hospital on a single day during a two-week period in June 2011 and January 2012. All patients who were present in the participating hospitals at 08.00 on the days of the surveys (June 2011 and January 2012) were included in the study. All inpatients who were being treated with an antibiotic for any reason, either oral or parenteral, in the hospital at 08.00 on the day of each survey were identified. To investigate the risk factors for HAI, more specific data were collected from the patients present in the six wards selected in each hospital for this purpose.

The required data were collected by hospital staff through reviewing both patients’ case notes and prescribing charts, using the European Surveillance of Antibiotic Consumption (ESAC) audit tool. ESAC is an international data collection network that aims to improve antibiotic prescribing through collecting data on patterns of antibiotic prescribing utilising a standard method.⁶ In addition, a form to collect the data related to HAI risk factors was developed.

Clinical pharmacists carried out this latter survey on their respective wards in each hospital. During the day of each point-prevalence survey, clinical pharmacists reviewed the patient medical notes, and when information was not documented, it was sought from the doctors or ward nurses. To ensure consistency of data collection across the participating hospitals, a guideline document was distributed to all the participating hospitals with the audit tool. In addition, clinical pharmacists, who were responsible for data collection, met in order to clarify how the data should be recorded.

Data collected included details on the number of inpatients in each department, patient age and gender, details of antibiotic agents used (dose per administration, number of doses per day and route of administration), duration of surgical prophylactic antibacterial use (coded as 1 dose, 1 day, >1 day) and anatomical site of the infection being treated. Antibiotics were grouped according to the Anatomic Therapeutic Chemical (ATC) classification (www.whocc.no/atc_ddd_index).

Compliance with the local hospital antibiotic guidelines and indication for treatment (that is, community-acquired infection, HAI or medical/surgical prophylaxis) were also documented. To facilitate comparison of patient groups across the study sites, the specialty (medicine, surgery, intensive care and other departments) was recorded for each patient. For inpatients admitted to the six wards, selected in each hospital for risk factor analysis, more specific data were collected.

The data included duration of hospital stay (from admission to day of survey), admission source (home, other hospital/department) and existence of indwelling devices during the current hospital stay (present at the time of the survey or during the seven days before the survey). These indwelling devices were categorised as: mechanical ventilation devices; central and peripheral venous catheters; urinary catheters; arterial catheters; dialysis catheters; wound drains; and nasogastric tubes.

HAIs were defined as cases where symptoms started ≥48h after admission to hospital.^25,26 The HAIs were classified into five categories as follows: (i) postoperative infection (within 30 days after surgery or one year after implant surgery); (ii) other intervention-related infections (intravenous catheter, vascular access port, continuous ambulatory peritoneal dialysis); (iii) Clostridium difficile-associated diarrhoea >48h after admission or <30 days after previous admission; (iv) other HAI; and (v), infection present on admission from another hospital.

Community-acquired infections were defined as cases where symptoms or antibiotic use started less than 48h after an admission to hospital. Assessment of compliance with the hospital antibiotic prescribing guidelines was performed by answering ‘yes’, ‘no’ or ‘not assessable’, that is, where the diagnosis was unclear, or ‘no information’. Patients who received more than one antibiotic were considered as being on combination treatment.

The assessment of patient therapy, as compliant or non-compliant with prescribing guidelines, was based on reference to individual hospital guidelines that were available on each ward of the study hospitals and on the Trusts’ intranets. These guidelines provided advice on diagnosis, dosing, duration and frequency of the antibiotic treatment, and also recommended the empirical therapy for the treatment of common infections, for example, urinary tract infection, respiratory tract infection, intra-abdominal infection and sepsis syndromes.

The prevalence of antibiotic prescription was determined as the percentage of the number of patients receiving any antibiotics out of the total number of patients present as inpatients in the study hospital, on the day of the survey. Prevalence of HAIs was determined by calculating the number of patients with HAI divided by the total number of patients in each hospital (expressed as a percentage) on the day of survey.

Risk factors for HAI were recorded in order to assess their relationship with the presence of HAI, included: Number of males/females, number of patients in different age groups, that is, ≤15 years, 16–40 years, 41–60 years and >60 years;²⁷ number of patients in different length of hospital stay groups, that is, ≤ 5days, 6-10 days and > 10 day;²² hospital ward type, and source of admission and existence of indwelling devices.

Descriptive statistics were used and results were expressed using standard statistical methods such as median, interquartile range, frequency (%), and 95% confidence intervals (CI). Appropriate univariate tests were used depending on the characteristics of data under analysis. The Mann–Whitney U-test was used with continuous variables (age, duration of hospital stay), while the Pearson Chi square (X2) test was used with the following categorical variables: age groups; gender; duration of hospital stay groups; admission source (home, other hospital/department); hospital ward type (medicine, surgery, intensive care unit, other wards); and existence of indwelling devices (mechanical ventilation devices, central and peripheral venous catheters, urinary catheters, arterial catheters, dialysis catheters, wound drains and nasogastric tubes).

The Fisher’s exact test was employed when > 20% of the expected frequency was <5. Variables with a p-value less than 0.25 in univariate analysis were included in a backward logistic regression analysis. Odds ratios (OR) with 95% CIs were used as a measure of the relative impact of each predictor on the outcome variable. The statistical analyses were performed on SPSS for Windows (Advanced Statistics Release 20.0, SPSS^® Inc, Chicago).

A total of 3070 inpatients were surveyed. Of these, 1070 (34.8%) were receiving an antibiotic. Most of the patients were present in medical departments, during both June 2011 (56.8%) and January 2012 (59.8%), followed by surgery (June 2011, 28.3%; January 2012, 27.7%), other department (June 2011 13.8%; January 2012, 10.3%) and intensive care unit (June 2011, 1.1%; January 2012, 2.2%). General characteristics of the study population are summarised in Table 1. The overall antibiotic prescription rate was 34.8% (1070 of 3070 patients). Out of 1363 patients (June 2011), 475 patients were treated with antibiotics (prevalence rate 34.8%), while in January 2012, the antibiotic prescribing prevalence rate was 34.9% (595/1707 patients). There was no statistically significant difference, with regards to the prevalence of antibiotic prescribing, between the two surveys (Pearson X2 test, p=0.999).

Different trends in antibiotic use were found during the two point-prevalence surveys. In June 2011, 666 antibiotic treatments were prescribed to 475 treated patients, compared with 867 antibiotic treatments prescribed to 595 treated patients in January 2012. The most frequently prescribed antibiotics during the two point-prevalence surveys were combinations of penicillins (32.6%, June 2011; 30.6%, January 2012).

This involved the use of amoxicillin-clavulanic acid (19.5% in June 2011; 12.5% in January 2012) and piperacillin-tazobactam (13.1% in June 2011; 18.1% in January 2012). Other most commonly prescribed antibiotic agents during the first point-prevalence survey were: penicillins with extended spectrum (11.1%); macrolides (9%); imidazole derivatives (7.4%); aminoglycosides (5.9%); beta lactamase-resistant penicillins (5.3%); and glycopeptide antibacterials (4.7%). During the second point-prevalence survey in January 2012, other antibiotics were: macrolides (9.5%); penicillins with extended spectrum (8.4%); imidazole derivatives (6.6%); aminoglycosides (6.2%); beta-lactamase resistant penicillins (6.5%); and glycopeptide antibacterial (Tables 2 and 3).

The most common indications for the prescribed antibiotics were community-acquired infection (66.2% (441/666) in June 2011; 59.9% (519/867) in January 2012), followed by HAI (22.3% (149/666) in June 2011; 25.9% (225/867) in January 2012), surgical prophylaxis (7.2% (48/666) in June 2011; 9.2% (80/867) in January 2012) and medical prophylaxis (4.2% (28/666) in June 2011; 5% (43/867) in January 2012]. The indications for which antibiotics were prescribed, for each hospital, are shown in Table 4.

The HAI category included five sub-categories, of which the highest antibiotic use was observed in the other HAI subgroup (64.4% (96/149) in June 2011; 56.8% (128/225) in January 2012), followed by, in descending order: postoperative infection subgroup (22.1% (33/149) in June 2011; 25.7% (58/225) in January 2012); infection present on admission from another hospital subgroup (4.7% (7/149) in June 2011; 7.1% (16/225) in January 2012); other intervention-related infections subgroup (4.0% (6/149) in June 2011; 6.2% (14/225) in January 2012); and C. difficile-associated diarrhoea subgroup (4.7% (7/149) in June 2011; 4.0% (9/225) in January 2012).

The most commonly prescribed antibiotic for community-acquired infection in June 2011 was amoxicillin-clavulanic acid (16.1%, 71/441) and piperacillin-tazobactam in January 2012 (18.5%, 96/519). Piperacillin-tazobactam was the most common antibiotic prescribed for HAI in both surveys (16.1% (24/149) in June 2011; 25.8% (58/225) in January 2012). Amoxicillin-clavulanic acid was the most commonly prescribed antibiotic for surgical prophylaxis in both surveys (75% (36/48) in June 2011; 36.3% (29/80) in January 2012).

Figure 1. Percentage compliance with hospital guidelines, for each hospital, during the two point-prevalence surveys in June 2011 and January 2012

The overall adherence rate to hospital guidelines was 66.6%. The results for the June 2011 survey showed that 63.2% (compliant, n=400; not compliant, n=105; not assessable, n=128) of the antibiotics prescribed were in compliance with the hospital guidelines, compared with 69.3% (compliant, n=573; not compliant, n=93; not assessable, n=161) in January 2012 (Figure 1). Antibiotic prescriptions that were not assessable were included in the overall denominator which resulted in a low adherence figure.

In the study sites, 98 patients received 128 antibiotic treatment courses for surgical prophylaxis during the two point-prevalence surveys, that is, 44 patients (single dose, n=25; one day, n=2; > 1 day, n=17; January 2011) and 54 patients (single dose, n=32; one day, n=7; > 1 day, n=15; January 2012). The overall duration of antibiotic treatments for surgical prophylaxis was greater than one day in 32.7% (32/98) of all patients who were receiving antibiotic prophylaxis treatment (38.6% in June 2011 compared with 27.8% in January 2012).

Table 5 shows details of duration of surgical prophylaxis for patients in the different study sites. The most commonly used antibiotic for surgical prophylaxis, for greater than one day’s duration during both periods of studies, was amoxicillin-clavulanic acid (70.0%, June 2011; 47.6%, January 2012).

The indication for the prescribed antibiotics was recorded in the patient notes for 82.7% (551/666) of the prescribed antibiotics (June 2011) compared with 90.5% (785/867; January 2012). Differences in the recording rates were observed in the study sites (Figure 2).

During the two point-prevalence survey periods, a total of 276 patients were recorded as having 282 HAIs (that is, three patients had more than one infection). The overall prevalence of HAI was 9.0% (276/3070), whereas the overall frequency of HAI was 9.2% (282/3070). The prevalence of HAIs increased from 7.9% (108/1363) in June 2011 to 9.8% (168/1707) in January 2012. No statistically significant differences in the proportions of HAI were found between the two surveys (p=0.065).

The HAI category included five sub-divisions, with the highest rate being found in ‘the other intervention-related infections’ subgroup (53.6%, 148/276), followed by, in descending order: ‘other hospital acquired infection’ subgroup (26%, 72/276); ‘postoperative infection’ subgroup (7.6%, 21/276); ‘infection present on admission from another hospital’ subgroup (6.5%, 18/276); and ‘C. difficile-associated diarrhoea’ subgroup (6.2%, 17/276).

Four hospitals (June 2011) and three hospitals (January 2012) participated in the study of risk factors of HAI, with each hospital involving six selected wards (intensive care unit, general medicine ward, general surgery ward, respiratory ward, care of the elderly ward and paediatric ward). On the days of the surveys, 738 patients were hospitalised in the selected wards in the participating hospitals.

Of the 738 patients, 83 had a HAI. All the variables (for which data were collected) that theoretically could be associated with HAIs were initially included in univariate analyses to examine the individual relationships between these variables and the rate of HAI. Seven variables had relationships with HAI in the univariate analysis, with p≤0.25 (that is, age group 16–40 (p=0.086) and age group >60 (p=0.042), length of stay >10 days (p=<0.001), source of admission: other hospital (p=0.213) and source of admission: other department (p=0.213), peripheral intravenous catheter (p<0.001), urinary catheter (p=0.034), arterial catheter (p=0.013), and dialysis catheter (p=0.224)).

In the stepwise backward logistic regression, involving the previous seven variables versus the HAI, variables found to be significantly and independently associated with HAI were: hospital stay more than ten days; being admitted from another hospital; having a peripheral intravenous catheter; and having an arterial catheter. Furthermore, according to the OR values, the odds of prevalence of HAI increased by a factor of 3.9 when the length of stay was more than ten days (OR 3.9, 95% CI 2.07–7.28, p<0.001), by a factor of 3.4 if the patient was admitted from another hospital (OR 3.4, 95% CI 1.56–7.28, p=0.002), by a factor of 3.7 if the patient had a peripheral intravenous catheter (OR 3.7, 95% CI 1.94–7.17, p<0.001) and by a factor of 4.6 if the patient had an arterial catheter (OR 4.6, 95% CI 1.97–10.50, p<0.001; Table 6).

The emergence and development of antibiotic-resistant organisms has become a serious problem worldwide.²⁸ The misuse of antibiotics plays a significant role in the spread of antibiotic resistance.⁵ Therefore, investigation of antibiotic use in hospitals is important to identify areas for improvement. The main aims of the current study were to assess current patterns of antibiotic prescribing in N. Ireland, as well as to evaluate potential risk factors for HAI within this environment. The method used in the present study was the point-prevalence survey as approach, which is inexpensive to perform and can provide very useful data.²⁹

In this study, the percentage antibiotic use (34.8%) was higher than that reported in Norwegian teaching hospitals (16.6%).³⁰ However, it was similar to other reported studies,^31,32 and lower than the rate of antibiotic use that has been reported in the USA (46.0%)³³ and Greece (51.4%).³⁴ Interestingly, the rate of antibiotic use in this study was similar to rates that were reported in previous studies in 2006 and 2009 in acute hospitals across Northern Ireland, that is, 34.2% and 32.0%, respectively.^35,36 High variability in antibiotic prescribing, between the participating hospitals, was observed. This finding could be due either to different local prescribing guidelines or cultures in relation to the treatment of infections or due to different case-mix of patients in the participating hospitals.

In the present study, the use of fluoroquinolones, second-generation cephalosporins and third-generation cephalosporins was low because these items were restricted following a major C. difficile infection (CDI) outbreak in a hospital in N. Ireland.³⁷ Amoxicillin-clavulanic acid and piperacillin-tazobactam were the most commonly prescribed individual antibiotics. Also these antibiotics were the most frequently prescribed antibiotics for the treatment of respiratory tract infection, the most common site of infection in the present study. According to antibiotic guidelines across N. Ireland, amoxicillin-clavulanic acid and piperacillin-tazobactam are recommended as first-line therapy to treat severe community-acquired pneumonia and hospital-acquired pneumonia, respectively.

Up to 87.0% of all antibiotics were used for the purpose of treatment (rather than prophylaxis), with almost two-thirds for community-acquired infections and one-third for HAI. A similar pattern of treatment versus prophylaxis has been observed in other studies conducted at university hospitals in five European countries (Croatia, Estonia, Latvia, Lithuania and Sweden).³⁸ In the present study, several targets for quality improvement in N. Ireland were identified: compliance with guidelines; duration of surgical prophylaxis; and recording the indication for treatment in patient notes. Compliance with guidelines, as an indicator, can be used in any quality improvement area.²⁴ It has been suggested that compliance with hospital antibiotic guidelines should not be <95%.³⁹ In the present study, the overall compliance rate to hospital guidelines (66.6%) was relatively low.

However, the compliance with hospital guidelines was better than that found in a range of other studies that used similar survey methods. For example, in the ESAC survey, the adherence rate was 62%;⁴⁰ another European study has reported adherence rates of 64.4%.⁴¹ Notably, in a recent previous study involving the same study sites, the overall compliance rate was 52.4%.³⁶ Moreover, the rate in the present study showed a marked improvement, from 63.2% in June 2011 to 69.3% in January 2012. Thus, continuous education interventions (developed and implemented by the antimicrobial management teams) and repeated antibiotic surveys may be an effective approach to improve staff adherence to hospital antibiotic guidelines.

Surgical prophylaxis is used to cover the pre-operative period in order to help prevent surgical site infections. The target duration of surgical prophylaxis should not exceed 24h.⁴¹ Usually one dose of antibiotic is sufficient.⁴⁰ Prolonged antibiotic prophylaxis has been shown to be no more efficacious, may increase antibiotic resistance and unnecessarily increases hospital cost.⁴² In the present study, the overall proportion of patients receiving surgical prophylaxis for more than one day (32.6%) was considerably higher than the recommended target (that is, should be 0%). The finding in this study is, however, lower than a number of other studies that used the same ESAC tool methodology, that is, 57.3% and 59.3% in the 2008 and 2009 ESAC surveys, respectively.^40,41 The findings suggest the need for intensive education programmes for surgeons that aim at increasing the adherence to surgical prophylaxis antibiotic guidelines.

Recording the indication for starting antibiotic treatment in patient notes has been used as a key performance indicator within the area of antibiotic prescribing; however, relatively few previous studies have documented this indicator.^40,41 The recommended standard for this indicator is that records should be present in 95% or more of cases.⁴¹ In the present study, the indication for treatment was documented in 87.0% of the studied cases. Although slightly lower than the recommended rate it was higher than reported in a number of other studies, for example, 59.4%²⁴ and 76.1%.⁴³

In the present study, the overall prevalence of HAI was 9.0%. This result is consistent with other reported findings internationally (ranging from 4.5% to 15.7%). ^35,44–46 The proportion of HAIs vary widely in the published literature, which can be explained by differences in the methodologies used, the definition of HAI, level of infection control programmes and patient populations studied.¹³ Several risk factors proved to be associated with HAI, in the present study. These factors have also been identified as risk factors in previous studies, that is, length of hospital stay,^22,25 patient admission from another hospital,⁴⁷ presence of a peripheral intravenous catheter^48,49 or arterial catheter.²¹ The identification of risk factors locally may help guide the infection control team in their efforts to improve patient care, through focusing on such patient groups.

The current study is associated with some limitations. Because the design of the study involved point-prevalence surveys and the data were gathered only for specific dates, data on the prescribing of antibiotics in periods before the actual audit days were not available. This meant that for HAI, it was impossible to differentiate between infections promoted by prior use of antibiotics or whether the antibiotics recorded were simply being used to treat HAI.

In this study, the appropriateness of antibiotic use was not assessed. However, markers of appropriate prescribing practice were addressed, for example, length of surgical prophylaxis. In addition, due to resource limitations, risk factor assessment was undertaken in only six wards in each hospital. The identified risk factors might, therefore, not represent the overall hospital. However, the selected wards were selected to be representative wards in the hospitals surveyed, and the main types of wards surveyed were represented in all hospitals (medical, surgical and ICU).

The following conclusions can be stated as a result of the present study: (i) using point-prevalence surveys, with standardised methodology (that is, ESAC audit tools), to evaluate antibiotic use and HAI in the present study sites was relatively inexpensive, feasible and not significantly time-consuming; (ii) the surveys provided valuable information which will act as a platform for further work on antibiotic use and resistance surveillance and to target quality improvement; (iii) the study demonstrated an improvement in adherence to hospital guidelines compared with previously conducted surveys; and (iv) length of hospital stay, admission sources and presence of peripheral intravenous or arterial catheters were significant predicators of HAI.

Overall, the study demonstrated the value of the point-prevalence survey approach in identifying possible targets for quality improvement, and as a means of monitoring and maintaining practice improvement.

• The misuse of antibiotics plays a significant role in the spread of antibiotic resistance; optimal prescribing of antibiotics is needed.
• The study identified several targets for quality improvement in N. Ireland: compliance with guidelines; duration of surgical prophylaxis; and recording the indication for treatment in patient notes.
• Several risk factors were identified to be associated with HAI, in the present study, including: length of hospital stay, patient admission from another hospital, presence of a peripheral intravenous, catheter or arterial catheter.
• The identification of risk factors locally may help guide the infection control team in their efforts to improve patient care, through focusing on such patient groups.
• Overall, the study demonstrated the value of the point-prevalence survey approach in identifying possible targets for quality improvement, and as a means of monitoring and maintaining practice improvement