Nur 410

2123

Age and Ageing 2021; 50: 2123–2132 https://doi.org/10.1093/ageing/afab169 Published electronically 28 August 2021

© The Author(s) 2021. Published by Oxford University Press on behalf of the British Geriatrics Society. All rights reserved. For permissions, please email: [email protected]

RESEARCH PAPER

Reduced Clostridioides difficile infections in hospitalised older people through multiple quality improvement strategies Carla Maria Dohrendorf1,2,†, Steffen Unkel3,†, Simone Scheithauer4, Martin Kaase4, Volker Meier5, Diana Fenz4, Jürgen Sasse6, Manfred Wappler7, Jutta Schweer-Herzig7, Tim Friede3, Utz Reichard8, Helmut Eiffert8, Roland Nau1,2, Jana Seele1,2

1Department of Geriatrics, Evangelisches Krankenhaus Göttingen-Weende, Göttingen, Germany 2Department of Neuropathology, University Medical Center Göttingen, Göttingen, Germany 3Department of Medical Statistics, University Medical Center Göttingen; Göttingen, Germany 4Institute for Infection Control and Infectious Diseases, University Medical Center Göttingen; Göttingen, Germany 5Hospital hygiene, Evangelisches Krankenhaus Göttingen-Weende; Göttingen, Germany 6Clinic for Geriatric Medicine, DRK Kliniken-Nordhessen; Kaufungen, Germany 7Clinic for Geriatric Medicine, Evangelisches Krankenhaus Gesundbrunnen Hofgeismar; Hofgeismar, Germany 8MVZ Wagnerstibbe for Medical Microbiology, Göttingen, Germany

Address correspondence to: Dr. Jana Seele, Department of Neuropathology, University Medical Center Göttingen, Göttingen, Germany & Department of Geriatrics, Evangelisches Krankenhaus Göttingen-Weende, Göttingen, Germany, Robert-Koch-Str. 40, 37075 Göttingen, Germany; Tel: +49-551-39-20489; Fax: +49-551-39-10800. Email: [email protected] †Contributed equally

Abstract Objectives: To reduce infections with Clostridioides difficile (CDI) in geriatric patients by interventions easily implementable in standard clinical care. Methods: Prevalence and incidence of CDI between January 2015 and February 2020 were analysed (n = 25,311 patients). Pre-intervention status was assessed from April 2016 to March 2017 (n = 4,922). Between May 2017 and August 2019, a monocentric interventional crossover study (n = 4,655) was conducted including standard care and three interventions: (A) sporicidal cleaning of hospital wards, (B) probiotics and (C) improvement in personal hygiene for CDI patients. This was followed by a multicentric comparison of the interventional bundle (A + B + C) between September 2019 and February 2020 (n = 2,593) with the pre-intervention phase. In 98 CDI cases and matched controls individual risk factors for the development of CDI were compared. Results: Time series analyses of CDI cases revealed a reduction in the prevalence of CDI in all three participating centres prior to the multicentric intervention phase. In the monocentric phase, no effect of individual interventions on CDI prevalence was identified. However, an aggregated analysis of CDI cases comparing the pre-intervention and the multicentric phase revealed a significant reduction in CDI prevalence. Risk factors for the development of CDI included use of antibiotics, anticoagulants, previous stay in long-term care facilities, prior hospital admissions, cardiac and renal failure, malnutrition and anaemia. Conclusions: The observed reduction in CDI may be attributed to heightened awareness of the study objectives and specific staff training. Individual interventions did not appear to reduce CDI prevalence. A further randomised trial would be necessary to confirm whether the bundle of interventions is truly effective.

Keywords: Clostridioides difficile infections,reduction of spores,probiotics, interventional study,geriatric patients,older people

Key Points • Reduction of infections with toxigenic Clostridioides difficile in geriatric patients. • Conduction of a monocentric interventional cross-over study followed by a multicentric evaluation of the interventional

bundle. • Combination of the reduction of spores in the environment and strengthening of the patients’ gut flora.

D ow

nloaded from https://academ

ic.oup.com /ageing/article/50/6/2123/6358411 by 81695661, O

U P on 17 N

ovem ber 2021

C. M. Dohrendorf et al.

Introduction

Infections with Clostridioides (C.) difficile (CDI) are among the leading causes of nosocomial infections [1]. Transmission occurs primarily by ingestion of C. difficile spores shed in vast quantities with the stools by infected and colonised indi- viduals. Spores can be found in the environment of affected patients and are easily distributed via the hands of healthcare providers because of their resistance to conventional alcohol- based disinfectants [2].

Depending on immunocompromising factors, perturba- tions in the gut microbiome, the C. difficile strain involved and the ingested dose, an exposure to C. difficile can result in asymptomatic colonisation or lead to an infection, whose clinical presentation ranges from mild diarrhoea to toxic mega-colon, sepsis and death [3].

Geriatric inpatients are particularly vulnerable as many known risk factors for the development of CDI pertain to them: old age, exposure to antibiotics, long hospitalisation duration, stay in long-term care facilities and severe comorbidities such as chronic renal failure or malnutrition [4]. The risk of infection among persons ≥65 years was 8.65 times higher than the risk among patients <65 years (95% confidence interval [CI] 8.16 to 9.31). Over 80% of CDI deaths occur in patients ≥65 years [1, 5].

There are several different approaches to prevent CDI. One is to limit exposure to C. difficile spores, e.g. by routine use of sporicidal agents for surface cleaning or by a daily change of hospital bed linen [6, 7]. Despite current standard infection control measures that focus on the isolation of infected patients, C. difficile spores can still be found in ward environments [6–8]. They are at least in part presumed to stem from asymptomatic carriers. In a study based on a small number of patients, 84% of nosocomial CDI appeared to be caused by strains introduced by asymptomatic carriers [9]. Other preventive approaches concentrate on the role of the microbiome: probiotics have shown promising results in the prevention of CDI, especially in geriatric wards, with reductions in CDI incidence of 61–66% as reported in three meta-analyses [10–12].

In the present study, we aimed to develop and evaluate the effect of interventions to reduce infections with C. difficile that are easily implemented in geriatric standard hospital care. These included the reduction of spores in the ward environment, the use of probiotics and improvement in the personal hygiene of CDI patients.

Methods

Study design and setting

The incidence [(CDI cases/occupancy days)∗1,000] and prevalence [(CDI cases/total cases)∗100] of all CDI and of nosocomial CDI were monitored in the three participating centres [Geriatrische Klinik, Evangelisches Krankenhaus Göttingen-Weende (EKW), Klinik für Geriatrie, DRK

Kliniken-Nordhessen Kassel (DRKK), Klinik für Geriatrie, Evangelisches Krankenhaus Gesundbrunnen Hofgeismar (EKH)] from January 2015 to February 2020 (last month included). The development of symptoms of a CDI within 48 h of admission to a geriatric ward was defined as a case acquired outside the geriatric departments and >48 h after admission as a nosocomial case. From May 2017 to August 2019 a monocentric four-armed interventional effectiveness study with crossover design was conducted at the EKW. The following interventions were implemented in four different hospital wards of the EKW: (group A) cleaning of all surfaces in all patients‘ rooms (Cleanisept� Wipes Forte containing benzalkonium chloride and didecyldimethylammonium chloride, Dr. Schumacher, Malsfeld, Germany), all hand contact surfaces (Cleanisept� Wipes Forte) and floors (Ultrasol� active containing peracetic acid, Dr. Schumacher, Malsfeld, Germany) with a sporicidal disinfectant; (group B) daily provision of probiotics (Actimel� containing Streptococcus thermophilus, Lactobacillus bulgaricus and Lactobacillus casei, Danone Germany, München-Haar, Germany) for all patients; and (group C) improvement in personal hygiene of patients with symptomatic CDI including a sporicidal laundry service for the personal clothes and daily change of the bedding. These interventions were carried out in addition to the standard hospital care which included isolation of patients with a symptomatic infection and sporicidal cleaning of the infected patient’s room once per day. The fourth ward (group D, standard hospital care) served as a control. Every ward comprised 19–25 beds and went through every intervention for 6 months and 3 weeks with 1 week wash-out in between. The interventions were randomly allocated to the wards (Figure 1). In the multicentric phase, a bundle comprising all three interventions in addition to the standard hospital care was implemented in the geriatric wards of EKW (4), EKH (3) and DRKK (3) in a total of 294 beds from September 2019 to February 2020 (Figure 1).

Twice per week the study nurse and/or study coordinator spoke with the hospital staff (including nurses, physicians and cleaning staff) about the study concept and implementa- tion of interventions as well as to listen closely to their needs and concerns.

Endpoints of the monocentric phase were the influence of the interventions A, B and C compared to standard hospital care on the infection rate and the dependence of the CDI rate on (1) the immune status of the patient, (2) the duration of the hospital stay, (3) the medication, especially the use of antibiotics and (4) comorbidities and underlying diseases.

The endpoint of the multicentric phase was the CDI prevalence during the multicentric phase (September 2019– February 2020) compared to a 1-year interval prior to the start of the monocentric phase (April 2016–March 2017). All endpoints of both the monocentric and the multicentric phase were evaluated on total CDI and nosocomial CDI as defined previously.

2124

D ow

nloaded from https://academ

ic.oup.com /ageing/article/50/6/2123/6358411 by 81695661, O

U P on 17 N

ovem ber 2021

Reduced Clostridioides difficile infections in hospitalised older people

Figure 1. Study design and implemented interventions. (A) The incidence and prevalence of CDI was calculated from January 2015 to February 2020. The monocentric phase of the study was implemented from May 2017 to August 2019 in the geriatric centre of EKW. The interventions were carried out for 6 months and 3 weeks with 1 week wash-out following. Afterwards the interventions rotated on the wards. The standard hospital care served as control. During the multicentric phase from September 2019 to February 2020 the bundle of the interventions A, B and C in addition to the standard hospital care was implemented in the geriatric centres of EKW, EKH and DRKK. (B) The interventions served to (A) reduce C. difficile and its spores in the hospital environment, (B) strengthen the gut microbiome of all patients and (C) improve the personal hygiene of patients with a symptomatic CDI in addition to (D), the standard hospital care.

Participants

Pre-intervention status was assessed from April 2016 to March 2017 (4,922 patients). The monocentric phase included 4,655 inpatients admitted to the geriatric depart- ment of the EKW between May 2017 and September 2019; in the multicentric phase 2,593 inpatients in the geriatric departments of EKW, EKH and DRKK between September 2019 and February 2020 were analysed.

For change point analyses, CDI cases from January 2015 to February 2020 were analysed (25,311 patients).

Microbiology

Stool samples of symptomatic patients were analysed by an enzyme immunoassay (EIA) detecting Clostridoides-specific glutamate dehydrogenase (GDH) to prove the presence of C. difficile. Then, the bacterial DNA encoding Toxin B was amplified by polymerase chain reaction (PCR) to identify toxigenic strains.

Case definition and data collection

A CDI case was defined by (i) unformed stools and (ii) positive results for GDH EIA and Toxin B PCR. CDI cases were identified by the routine hospital infection surveillance systems. Only the patient’s first episode during the study period was recorded as a new case. CDI was considered severe, if the patient died, was transferred to an ICU or received colonic surgery with CDI as a plausible cause based on the medical records.

Data of CDI patients were compared with data of control patients by matched-pairs analysis. For the control group, patients without CDI or known C. difficile colonisation who

had stayed at the EKW during the monocentric phase for at least 2 days were selected in a 1:1 matching with respect to age and sex. Laboratory data, medication during hospital stay, demographics and comorbidities were collected using the medical records. Laboratory results were collected at admission to the geriatric ward (or from the sample closest to this time point; t0). As a second blood analysis in CDI cases the measurements obtained closest to the C. difficile-positive stool sampling were used (t1). For control patients, t1 was set at 12 days after admission (or the latest results available, if the stay was shorter than 12 days). This was based on a pre- evaluation of CDI patients from 2016 in which the median of the interval admission – CDI was 12 days.

Ethics

This study was approved by the Ethics Committee of the University Medical Centre Göttingen, Georg-August- University Göttingen (application number 22/1/17). Individual informed consent was waived since all measures implemented in this study aimed to maintain and improve the quality of patient care. Upon admission, all patients gave consent that their data can be used for research purposes by staff subjected to medical confidentiality. A data protection declaration was signed by all project members.

Statistical power calculation and data analysis

A full survey was carried out on all wards of the EKW (phase 1) and in all participating centres (phase 2). Statistical power calculation based on the pre-defined endpoints was performed prior to the beginning of the study under the following assumptions. Monocentric phase (phase 1): on average, 225 patients will be treated on each ward of the

2125

D ow

nloaded from https://academ

ic.oup.com /ageing/article/50/6/2123/6358411 by 81695661, O

U P on 17 N

ovem ber 2021

C. M. Dohrendorf et al.

EKW during one intervention period of 6 months plus 3 weeks (1 week wash-out period), i.e. after four rotations, 900 patients are treated with intervention A, B, C and stan- dard care (D). Compared with the pre-intervention status, the number of CDI cases will be reduced by each interven- tion (A, B, C) by 25%. These assumptions led to an estimated power of >80%. Multicentric phase (phase 2): a total of 2,700 patients will be treated in the centres participating in the 6-months period of the bundle intervention. Without interventions, a total number of 90 CDI cases were assumed to occur. The number of CDI cases will be reduced by 50% during the multicentric phase. These assumptions led to an estimated power of >95%.

Univariate analysis of individual risk factors was con- ducted that compared CDI patients with their matched controls, and a subgroup analysis was performed, including only nosocomial cases and their respective control patients. Dichotomous variables were compared by using Fisher’s exact test. Categorical variables (with more than two cat- egories on one variable) were compared using Wilcoxon’s signed rank test after being ranked according to the severity of the impairment of the function studied. Continuous variables were compared using Wilcoxon’s signed rank test for matched pairs, and if matching was not feasible by Mann–Whitney U test. Change point analyses were carried out to detect the point at which a structural break (change in the normally distributed mean) of the times series of monthly CDI prevalences and incidences occurred [13]. The results of the change point analyses were visualised in a similar fashion as in statistical process control, in which control charts are used to distinguish between common and special causes of variation. Control charts typically include a plot of the data over time, and one or more additional lines, which e.g. represent the means of the time series, or indicate when a signal of special cause variation has occurred [14].

A two-tailed P value ≤ 0.05 was considered statisti- cally significant. Statistical analyses were conducted using R version 4.0.1. [15] and GraphPad Prism, Version 6.01 (2012).

Results

Change point analyses of CDI prevalence and incidence

In the present study the effect of interventions to reduce infections with C. difficile that are easily implementable into geriatric standard hospital care were evaluated. These included (A) the reduction of spores in the ward, (B) the use of probiotics and (C) improvement in personal hygiene of CDI patients. The prevalence and incidence of CDI cases were analysed from January 2015 to February 2020, i.e. from 16 months prior to the start of the monocentric phase up to the end of the multicentric phase of the study. Change point analyses revealed that the prevalence of nosocomial and total (nosocomial, i.e. acquired in the geriatric department,

plus externally acquired) CDI cases significantly dropped at the Departments of Geriatrics of the EKW between March and April 2015, at the EKH between March and April 2017 and at the DRKK between April and May 2016 (Figure 2). A change point in regard to the incidence of CDI cases was only detectable for total CDI cases at the EKW (Supplementary Figure 1).

Influence of the individual interventions on CDI infection rate (monocentric phase)

During the monocentric phase of the study, the interven- tions A–D were carried out as individual interventions on four different geriatric wards of the EKW from May 2017 to August 2019 with a rotation of the interventions after 6 months and 3 weeks, each separated by 1 week of wash-out. An analysis of CDI cases at the end of the monocentric phase showed that none of the experimental interventions A, B or C (total CDI incidence A: 1.37, B: 1.99, C: 1.33; nosocomial CDI incidence A: 1.20, B: 1.29, C: 1.04) reduced the CDI infection rate compared to standard hospital care D (total CDI incidence 0.86, nosocomial CDI incidence 0.59). Because of differences in the composition of the patients treated on the individual wards, their CDI incidence was not equal. Total CDI incidence on the individual wards ranged from 1.02 to 1.97, and nosocomial CDI incidence from 0.91 to 1.42 illustrating the necessity of the chosen study design (allocation of the sequence of interventions to the wards randomly, every ward went through every intervention).

Individual risk factors for the development of CDI

During the monocentric phase of the study, data of 98 matched pairs (CDI vs. control) were analysed for individual risk factors for the development of CDI (Tables 1 and 2, Supplementary Table 1). The median age in both groups was 82 years. Of the 98 CDI cases, 88.8% were nosocomial infections, 11.2% were recurrent episodes and 4.3% met criteria of severe CDI. Median length of stay at the geriatric ward was 6.5 days longer for CDI patients, and they were discharged with a worse outcome. CDI was significantly associated with prior hospital stays, previous residency in a long-term care facility, invasive feeding and worse scores in geriatric assessments including the Charlson Comorbidity Index (CCI). CDI patients suffered more often from cardiac failure, malnutrition, anaemia, hypothyroidism, renal failure and dialysis, while Parkinson’s disease and conservatively treated fractures were more frequent in the control group.

CDI patients showed higher levels of serum immunoglob- ulin A (IgA), infection parameters (both at time of admission and sampling) and plasma alanine transaminase, whereas total plasma calcium was lower as a consequence of lower total protein and albumin concentrations. When plasma calcium was corrected for albumin using Payne’s formula [16], calcium was not lower in CDI than in the respective control patients. Plasma potassium levels fell in CDI patients

2126

D ow

nloaded from https://academ

ic.oup.com /ageing/article/50/6/2123/6358411 by 81695661, O

U P on 17 N

ovem ber 2021

Reduced Clostridioides difficile infections in hospitalised older people

Ta bl

e 1.

C ha

ra ct

er ist

ic so

fC D

Ic as

es an

d co

nt ro

ls Va

ria bl

e C

D I,

m ed

ia n

(I Q

R )

(N =

98 )

co nt

ro ls,

m ed

ia n

(I Q

R )

(N =

98 )

O R

P va

lu e

N os

oc om

ia lC

D I,

m ed

ia n

(I Q

R )

(N =

87 )

co nt

ro ls,

m ed

ia n

(I Q

R )

(N =

87 )

O R

P va

lu e

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

. .

Le ng

th of

sta y

in ge

ria tr

ic w

ar d

[d ay

s] 22

.5 (1

6; 35

.2 5)

16 (1

4; 20

) <

0. 00

01 †

27 (1

7; 36

) 16

(1 4;

20 )

< 0.

00 01

†

Pr io

rh os

pi ta

la dm

iss io

n in

th e

la st

4 w

ee ks

,N o.

(% )

72 (7

3. 5%

) 56

(5 7.

1% )

2. 08

0. 02

4∗ 63

(7 2.

4% )

50 (5

7. 5%

) 1.

94 0.

05 6∗

C C

I‡ 4

(2 ;5

) 2

(1 ;4

) 0.

00 02

† 4

(2 ;5

) 2

(1 ;4

) 0.

00 3†

C ar

di ac

in su

ffi ci

en cy

,N o.

(% )

54 (5

5% )

31 (3

2% )

2. 65

0. 00

15 ∗

46 (5

3% )

29 (3

3% )

2. 24

0. 01

4∗ D

ia ly

sis ,N

o. (%

) 12

(1 2%

) 0

(0 %

) n.

a. 0.

00 03

∗ 11

(1 3%

) 0

(0 %

) n.

a. 0.

00 07

∗ Pa

rk in

so n’

sd ise

as e,

N o.

(% )

11 (1

1% )

34 (3

5% )

0. 24

0. 00

01 ∗

9 (1

0% )

32 (3

7% )

0. 2

< 0.

00 01

∗ Fr

ac tu

re (c

on se

rv at

iv e

tre at

m en

t), N

o. (%

) 4

(4 %

) 16

(1 6%

) 0.

22 0.

00 8∗

4 (5

% )

13 (1

5% )

0. 27

0. 03

8∗ Fe

ed in

g, N

o. (%

) or

al 83

(8 4.

7% )

94 (9

5. 9%

) 0.

02 8†

73 (8

3. 9%

) 83

(9 5.

4% )

0. 04

6†

na so

ga str

ic tu

be 1

(1 %

) 0

(0 %

) 1

(1 .1

% )

0 (0

% )

PE G

2 (2

% )

0 (0

% )

2 (2

.3 %

) 0

(0 %

) pa

re nt

er al

12 (1

2. 2%

) 4

(4 .1

% )

11 (1

2. 6%

) 4

(4 .6

% )

Ty pe

of re

sid en

ce ,N

o. (%

) pr

iv at

e 45

(4 6.

4% )

56 (5

8. 3%

) 0.

01 8†

43 (5

0% )

48 (5

6. 5%

) 0.

10 3†

pr iv

at e

w ith

ca re

se rv

ic e

26 (2

6. 8%

) 27

(2 8.

1% )

23 (2

6. 7%

) 27

(3 1.

8% )

lo ng

-te rm

ca re

fa ci

lit y

26 (2

6. 8%

) 13

(1 3.

5% )

20 (2

3. 3%

) 10

(1 1.

8% )

Th yr

oi d

sc or

e∗ ∗ ,

m ea

n (I

Q R

) 3.

23 (3

;3 )

2. 99

(3 ;3

) 0.

02 5†

3. 25

(3 ;3

) 3.

00 (3

;3 )

0. 03

8†

Ig G

[m g/

dl ]

90 6

(6 85

;1 ,1

37 )

88 1

(6 78

;1 ,0

92 )

0. 75

† 90

4 (6

85 ;1

,1 01

) 88

7 (6

76 ;1

,1 01

) 0.

62 †

Ig M

[m g/

dl ]

73 (5

0; 11

0) 61

(3 8.

5; 92

.5 )

0. 08

7† 73

(5 0.

5; 11

0) 59

(3 8;

92 )

0. 06

7†

Ig A

[m g/

dl ]

28 1

(1 94

;3 72

) 21

2 (1

56 ;3

17 )

0. 02

6† 27

6 (1

97 ;3

71 .5

) 20

9 (1

55 ;3

21 )

0. 01

6†

To ta

lp ro

te in

[g /d

l] 6.

3 (5

.6 ;6

.7 5)

6. 6

(6 .2

;7 .1

) 0.

00 5†

6. 35

(5 .6

;6 .7

3) 6.

55 (6

.2 ;7

.1 )

0. 01

6†

Al bu

m in

[g /l]

26 .0

5 (2

3. 13

;2 9.

73 )

31 .8

(2 7.

15 ;3

5. 5)

< 0.

00 01

† 26

.2 (2

3. 4;

29 .8

) 31

.5 (2

6. 93

;3 5.

03 )

< 0.

00 01

†

C re

at in

in e

t 0 [m

g/ dl

] 1.

14 (0

.8 3;

1. 65

) 0.

97 (0

.7 7;

1. 19

) 0.

00 08

† 1.

14 (0

.8 3;

1. 63

) 0.

96 (0

.7 8;

1. 2)

0. 00

2†

H ae

m og

lo bi

n t 0

[g /d

l] 10

.3 (8

.9 ;1

1. 55

) 11

.5 (9

.5 ;1

3. 1)

0. 00

05 †

10 .2

(8 .7

;1 1.

5) 11

.4 (9

.5 ;1

3. 1)

0. 00

1†

Le uk

oc yt

es t 0

[1 03 /μ

l] 9.

5 (7

.1 5;

12 .7

4) 8.

07 (6

.8 2;

9. 57

) 0.

00 07

† 9.

26 (7

.1 3;

12 .2

6) 8.

29 (6

.9 5;

9. 6)

0. 00

4†

Le uk

oc yt

es t 1

[1 03 /μ

l] 9.

30 5

(6 .9

8; 12

.5 2)

6. 56

(5 .4

0; 8.

16 )

< 0.

00 01

† 8.

96 (6

.8 ;1

1. 95

) 6.

89 5

(5 .4

5; 8.

17 )

< 0.

00 01

†

O R

= O

dd s

ra tio

,I Q

R =

In te

rq ua

rt ile

ra ng

e, C

C I=

C ha

rls on

co m

or bi

di ty

in de

x, PE

G =

Pe rc

ut an

eo us

en do

sc op

ic ga

str os

to m

y t 0

= tim

e of

ad m

iss io

n to

ge ria

tr ic

w ar

d, t 1

= tim

e of

T C

D -p

os iti

ve sa

m pl

e co

lle ct

io n

(c on

tro ls:

12 da

ys af

te ra

dm iss

io n

to th

e ge

ria tr

ic w

ar d

or tim

e of

di sc

ha rg

e fro

m ge

ria tr

ic w

ar d

if th

e sta

y w

as sh

or te

rt ha

n 12

da ys

), �

= t 1

– t 0

∗ P va

lu es

ca lc

ul at

ed w

ith Fi

sh er

ex ac

tt es

t† P

va lu

es ca

lc ul

at ed

w ith

W ilc

ox on

sig ne

d- ra

nk te

st ‡

C C

Ic al

cu la

te d

w ith

ou tc

on sid

er at

io n

of th

ea ge

∗∗ C

la ss

ifi ca

tio n

of th

yr oi

d sta

tu s:

1 =

ov er

th yp

er th

yr oi

di sm

,2 =

su bc

lin ic

al hy

pe rt

hy ro

id ism

,3 =

eu th

yr eo

sis ,4

= su

bc lin

ic al

hy po

th yr

oi di

sm ,

no tr

eq ui

rin g

tre at

m en

t, 5 =

su bc

lin ic

al hy

po th

yr oi

di sm

,r eq

ui rin

g tre

at m

en t,

6 =

ov er

th yp

ot hy

ro id

ism

2127

D ow

nloaded from https://academ

ic.oup.com /ageing/article/50/6/2123/6358411 by 81695661, O

U P on 17 N

ovem ber 2021

C. M. Dohrendorf et al.

Figure 2. Change point analysis of nosocomial (A, B, C) and total (D, E, F) CDI prevalence from January 2015 (month 0) to February 2020 (month 60) in the geriatric centres of EKW (A, D), EKH (B, E) and DRKK (C, F). The prevalence per month was calculated as CDI cases/total cases∗100.

from admission to time of stool sampling, unsurprisingly during a diarrhoeal illness.

Medication associated with CDI were anticoagulants, loop diuretics, antiepileptics, antimotility agents, and signif- icant only for nosocomial CDI, benzodiazepines. Calcium channel blockers were prescribed more often in the control group. The use of proton pump inhibitors and immuno- suppressants was not significantly different. The number of total prescriptions during the geriatric stay was higher in the nosocomial CDI group.

Further risk factors for development of CDI included the use of antibiotics, higher numbers of different antibiotics and

more days of antibiotic use. Odds ratios (OR) were similar for the different routes of administration.

Influence of the intervention bundle on CDI prevalence (multicentric phase)

As individual interventions during the monocentric phase did not reduce the incidence of CDI, we decided to com- bine all interventions in addition to the standard hospital care during the multicentric phase in the three participat- ing geriatric centres for 6 months. An aggregated analysis of the nosocomial and total (nosocomial plus ambulant)

2128

D ow

nloaded from https://academ

ic.oup.com /ageing/article/50/6/2123/6358411 by 81695661, O

U P on 17 N

ovem ber 2021

Reduced Clostridioides difficile infections in hospitalised older people

Table 2. Use of antibiotics and other medication in CDI cases and controls CDI vs. controls Nosocomial CDI vs. controls

Variable CDI, No. (%) (N = 98)

controls, No. (%) (N = 98)

OR P value Nosocomial CDI, No. (%) (N = 87)

controls, No. (%) (N = 87)

OR P value

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Antibiotics 74 (76%) 50 (51%) 2.96 0.0006∗ 68 (78%) 46 (53%) 3.19 0.0007∗ Aminopenicillins 48 (49%) 23 (23%) 3.13 0.0003∗ 43 (49%) 21 (24%) 3.07 0.0009∗ Piperacillin/tazobactam 32 (33%) 12 (12%) 3.47 0.001∗ 31 (36%) 11 (13%) 3.82 0.0006∗ Linezolid 6 (6%) 0 (0%) n.a. 0.029∗ 6 (7%) 0 (0%) n.a. 0.029∗ Days with antibiotic use, median (IQR)

6 (0.75; 12) 1 (0; 6) <0.0001† 8 (1; 12) 1 (0; 6) <0.0001†

Number of antibiotics used, median (IQR)

1 (0.75; 3) 1 (0; 1.25) <0.0001† 2 (1; 3) 1 (0; 2) <0.0001†

Anticoagulants 57 (58%) 35 (36%) 2.50 0.003∗ 55 (63%) 35 (40%) 2.55 0.004∗ low molecular weight heparin

19 (19%) 7 (7%) 3.13 0.019∗ 19 (22%) 7 (8%) 3.19 0.018∗

unfractionated heparin 8 (8%) 1 (1%) 8.62 0.035∗ 8 (9%) 1 (1%) 8.71 0.034∗ Loop diuretics 61 (62%) 48 (49%) 1.72 0.084∗ 58 (67%) 44 (51%) 1.95 0.045∗ Calcium channel blockers 25 (26%) 41 (42%) 0.48 0.023∗ 24 (28%) 38 (44%) 0.49 0.039∗ Benzodiazepines 14 (14%) 7 (7%) 2.17 0.165∗ 14 (16%) 4 (5%) 3.98 0.023∗ Antiepileptics 23 (23%) 11 (11%) 2.43 0.037∗ 20 (23%) 10 (11%) 2.30 0.070∗ Number of total prescriptions, median (IQR)

9 (7; 12) 9 (6; 11) 0.145 10 (7; 12) 9 (6; 11) 0.029†

OR = Odds ratio, IQR = Interquartile range ∗P values calculated with Fisher exact test †P values calculated with Wilcoxon signed-rank test

CDI cases of all three participating centres comparing the 1-year interval before the start of the study (April 2016 until March 2017) with the 6-month interval of the mul- ticentric phase (from September 2019 to February 2020) revealed a significant reduction in CDI cases during the study (nosocomial: OR 0.60, 95% CI 0.40–0.90; total: OR 0.56, 95% CI 0.39–0.81; Figure 3). When analysing each centre separately, the reduction in the odds of infection were statistically significant in EKH (nosocomial: OR 0.53, 95% CI 0.25–1.11; total: OR 0.40, 95% CI 0.20–0.79) and DRKK (nosocomial: OR 0.39, 95% CI 0.16–0.95; total: OR 0.41, 95% CI 0.18–0.94), whereas in EKW it failed to reach statistical significance (nosocomial: OR 0.83, 95% CI 0.46–1.50; total: OR 0.84, 95% CI 0.49–1.42; Figure 3).

Discussion

In this study, change point analyses and pre-to-post inter- vention comparison revealed that the total number as well as number of nosocomial CDI cases declined in the partic- ipating centres. No single intervention in the monocentric phase was found to be effective, meaning that in respect to the a priori defined endpoint of the monocentric phase of this study, each intervention by itself was not found to reduce CDI prevalence. However, the bundle of interventions in combination with the associated staff awareness and educa- tion appeared to be effective in the multicentric prior-post comparison.

The change point analyses suggested that the heightened awareness through the study objectives accompanied by discussions and staff training was key to the reduction in

Figure 3. Analysis of nosocomial and total CDI cases in the three participating centres. The CDI prevalence prior the study (April 2016–March 2017) was compared with the CDI preva- lence during the multicentric phase of the study (09/2019– 02/2020). The Odds ratio was calculated for each centre and in an aggregated analysis for all centres. Odds ratio and 95% confidence intervals are shown.

CDI incidence. This is similar to a situation in Great Britain in 2008, where as a consequence of the high incidence of CDI, a national target was set for a 30% reduction in CDI by 2010–11, which was achieved in spite of very heterogeneous approaches towards CDI prevention [17].

Conceivable reasons for the failure of the monocentric phase of this study were (a) the lack of effectivity of the

2129

D ow

nloaded from https://academ

ic.oup.com /ageing/article/50/6/2123/6358411 by 81695661, O

U P on 17 N

ovem ber 2021

C. M. Dohrendorf et al.

individual interventions, or (b) insufficient adherence to the intervention protocols by the participating staff or patients, or both.

As to (a): The evidence for the effectivity of some indi- vidual interventions was fairly low: chlorine or oxygen-based daily and terminal cleaning to reduce the concentration of C. difficile spores was of all interventions most likely to reduce CDI [18]. Installing a laundry service for infected patients was supported by only few observations [6, 7]. The use of probiotic yoghurt was supported by more studies [11, 19–22]. In studies with a baseline CDAD risk of 0 to 2% and 3 to 5%, probiotics were found to be ineffective, but studies including patients with a baseline risk of >5% for developing CDAD showed a risk reduction of approximately 70% [11]. In all participating centres the baseline risk was <5%. In regard to the probiotics, the effect was not always reproducible [23], and as a consequence of strong economic interests a publication bias in favour of positive studies is suspected.

(b) The staff compliance varied: achieving compliance with appropriate cleaning technique was difficult, because depending on the institution and also on the composition of the flooring, staff and patients complained of bad odour and mucosal irritation. As a consequence, the areas cleaned by chlorine bleach had to be reduced in all centres during the multicentric study. Probiotic yoghurt was not distributed daily on all wards as a consequence of work overload or prejudices on the part of the staff in regard to the inter- ventions. Moreover, the laundry service was not propagated rigorously enough. The compliance of the patients depended on this measure: The laundry service was rejected by many patients because of the fear of losing their clothing. In total, in EKW 10% of the patients accepted the laundry service. In EKH and DRKK, it was not accepted by the patients. The daily distribution of probiotic yoghurt was the measure most readily accepted by all patients. On an average, 81% of the patients received and accepted the probiotic yoghurt in all participating centres. The supply of a probiotic yoghurt will be continued in EKW. Especially for geriatric patients who often suffer from swallowing disorders, yoghurt is an appropriate snack.

Change point analyses of the entire study period revealed that all centres reduced the prevalence of CDI infections. Bundled interventions of different compositions are con- sidered highly effective [18]. After we had noticed that no individual measure appeared to be effective in the monocen- tric study, we decided to use all interventions as a bundle. The bundle of interventions was effective, when all centres were analysed together, and in two of three centres, when each centre was analysed separately. This discrepancy is a consequence of the different change points: in EKW the change point already occurred in March 2015, i.e. prior to the control period of the multicentric study, whereas in EKH and DRKK the change points occurred during the control period of the multicentric study. No further change point occurred at the beginning of the multicentric phase, i.e. the bundle intervention. Due to the different times when the

change occurred, we hypothesise that the time of the change point depended on the awareness of the problem of frequent CDI infections by the heads of the institutions and the educational measures taken thereafter. In EKW, the change point corresponded to the start of the planning of this project including discussions among the staff about correct hygiene measures. In DRKK the change point was associated with efforts of the chief physician (J.Sa.) to implement rational antibiotic therapy. In EKH, the change point was close to joint meetings in preparation of the multicentric phase of this study. In particular, the colleagues of this institution were impressed by a lecture of M.K. on the pathophysiol- ogy, molecular biology and epidemiology of CDI. In this respect, the results of the present study resemble experiences published concerning the prevention of overwhelming post- splenectomy infection (OPSI) where a bundle of measures involving vaccination, antibiotic prophylaxis and patient education can prevent infections, although the effectiveness of some of the individual interventions remains unclear [24].

In general, CDI incidences and prevalences were com- parable to other clinical studies [25]. In the present study, CDI was defined by unformed stools and positive results for GDH EIA and Toxin B PCR. In older patients, infections may present with mild symptoms, which may have led to a slight underestimation of the true incidence [2]. Individual risk factors of CDI were similar in the present study as in previous investigations: We confirmed the main risk fac- tors, previous use of antibiotics, renal failure, malnutrition, but not previous ICU stay, proton pump inhibitor and immunosuppressant treatment. Since we matched for age, we could not reproduce this well-known fact [26]. The effect of stay in long-term care facilities, lower functional status and comorbidities may be mediated by several factors. Frequent contact with healthcare facilities entails increased exposure to C. difficile spores [27]. Impairment of the immune sys- tem (e.g. by malnutrition or renal failure) facilitates the development of CDI directly or may result in other infec- tions demanding antibiotic therapy thereby disbalancing the gut microbiome. Anticoagulation was associated with an increased risk of developing CDI and nosocomial CDI. Approximately 40% of the patients at our institution receive anticoagulants mostly due to nonvalvular atrial fibrillation or flutter (AF) [28]. AF is associated with a variety of diseases and with an increased mortality (odds ratio approx. 2.5) [29]. Therefore, we hypothesise that in the present study anticoagulated patients had a higher burden of diseases than patients receiving no anticoagulation, rendering them more susceptible to CDI. The median CCI of anticoagulated CDI- infected patients was 4, whereas the median CCI of CDI- infected not anticoagulated patients was 3, and the CCI of the respective control patients in both subgroups was 2.

Our study has several limitations. It consisted of an obser- vational and a randomised intervention part. In the ran- domised intervention part (monocentric phase), no reduc- tion in CDI was achieved. For the observational part, our study succeeded in reducing C. difficile infections in the par- ticipating departments. In line with the majority of the other

2130

D ow

nloaded from https://academ

ic.oup.com /ageing/article/50/6/2123/6358411 by 81695661, O

U P on 17 N

ovem ber 2021

Reduced Clostridioides difficile infections in hospitalised older people

studies on CDI in acute care hospitals, for the multicentric study we used a simple pre- and post-intervention design. The use of a step-wedge design might have improved the quality of the data [18].

Conclusion

In conclusion, the observed reduction in CDI may be attributed to a heightened awareness of the study objectives, as well as to specific staff training. The individual interven- tions did not appear to reduce CDI prevalence. The bundle of interventions and the accompanying staff training reduced the high incidence and prevalence of CDI by approx. 40%, whereas the individual measures appeared to be ineffective. The results of the present study could be characterised by the proverb ‘A danger foreseen is half avoided’ [30]. It remains open whether the bundle of interventions was truly effective or whether the reduction in CDI incidence primarily was achieved by an increased awareness of the problem and by training measures accompanying the implementation of this study. Therefore, the promising effects of the combined intervention would need to be confirmed in a future cluster- randomised multicentre trial in which the wards represent the clusters.

Supplementary Data: Supplementary data mentioned in the text are available to subscribers in Age and Ageing online.

Acknowledgements: We thank Cynthia Bunker for project administration and careful language editing.

Declaration of Sources of Funding: This study was funded by Gemeinsamer Bundesausschuss, Innovationsausschuss, project number: 01VSF16059. The funding bodies did not influence the design of the study, collection, analysis, and interpretation of data and manuscript writing.

Declaration of Conflicts of Interest: SS and TF report other grants from Gemeinsamer Bundesausschuss (G- BA), Innovationsausschuss. TF received personal fees from Novartis, Bayer, Janssen, SGS, Roche, Boehringer Ingelheim, Daiichi-Sankyo, Galapagos, Penumbra, Parexel, Vifor, BiosenseWebster, CSL Behring, Fresenius Kabi, Coherex Medical, LivaNova, all outside the submitted work. RN received honoraria for lectures from Bayer Vital, Pfizer, Bristol-Myers Squibb and Desitin, and research support from Novartis, B. Braun Foundation, Deutsche Gesellschaft für Geriatrie, and Strathmann GmbH, all unrelated to the submitted work. The other authors declare that they have no conflicts of interest.

References

1. Lessa FC, Mu Y, Bamberg WM et al. Burden of Clostridium difficile Infection in the United States. N Engl J Med 2015; 372: 825–34.

2. Yoshikawa TT, Norman DC. Geriatric Infectious Diseases: Current Concepts on Diagnosis and Management. J Am Geriatr Soc 2017; 65: 631–41.

3. Leffler DA, Lamont JT. Clostridium difficile infection. N Engl J Med 2015; 372: 1539–48.

4. Jump RL. Clostridium difficile infection in older adults. Aging health 2013; 9: 403–14.

5. Balsells E, Shi T, Leese C et al. Global burden of Clostridium difficile infections: a systematic review and meta-analysis. J Glob Health 2019; 9: 10407.

6. Fekety R, Kim KH, Brown D et al. Epidemiology of Antibiotic-Associated Colitis. Isolation of Clostridium Dif- ficile from the Hospital Environment. Am J Med 1981; 70: 906–8.

7. Goldstein EJC, Johnson S, Maziade P-J et al. Pathway to Pre- vention of Nosocomial Clostridium difficile Infection. Clin Infect Dis 2015; 60: S148–58.

8. Dumford DM, Nerandzic MM, Eckstein BC et al. What is on that keyboard? Detecting hidden environmental reservoirs of Clostridium difficile during an outbreak associated with North American pulsed-field gel electrophoresis type 1 strains. Am J Infect Control 2009; 37: 15–9.

9. Clabots CR, Johnson S, Olson MM et al. Acquisition of Clostridium Difficile by Hospitalized Patients: Evidence for Colonized New Admissions as a Source of Infection. J Infect Dis 1992; 166: 561–7.

10. Johnston BC, Ma SSY, Goldenberg JZ et al. Probiotics for the Prevention of Clostridium difficile-Associated Diarrhea A Systematic Review and Meta-analysis. Ann Intern Med 2012; 157: 878–U225.

11. Goldenberg JZ, Ma SSY, Saxton JD et al. Probiotics for the prevention of Clostridium difficile-associated diarrhea in adults and children. Cochrane Database Syst Rev 2013; 5: 1–107.

12. Johnson S, Maziade JP, McFarland LV et al. Is primary pre- vention of Clostridium difficile infection possible with specific probiotics? Int J Infect Dis 2012; 16: e786–92.

13. Hinkley DV. Inference About the Change-Point in a Sequence of Random Variables. Biometrika 1970; 57: 1–17.

14. Mohammed MA, Worthington P, Woodall WH. Plotting basic control charts: tutorial notes for healthcare practitioners. Qual Saf Health Care 2008; 17: 137–45.

15. Core R. Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical. Computing, 2013.

16. Payne RB, Little AJ, Williams RB et al. Interpretation of serum calcium in patients with abnormal serum proteins. Br Med J 1973; 4: 643–6.

17. Duerden BI. Contribution of a government target to control- ling Clostridium difficile in the NHS in England. Anaerobe 2011; 17: 175–9.

18. Louh IK, Greendyke WG, Hermann EA et al. Clostrid- ium Difficile Infection in Acute Care Hospitals: Systematic Review and Best Practices for Prevention. Infect Control Hosp Epidemiol 2017; 38: 476–82.

19. McFarland LV, Johnson SB, Evans CT. Perils and pitfalls of probiotic quasi-experimental studies for primary prevention of Clostridioides difficile infection: A review of the evidence. Am J Infect Control 2020 49: 375–84.

20. Mills JP, Rao K, Young VB. Probiotics for prevention of Clostridium difficile infection. Curr Opin Gastroenterol 2018; 34: 3–10.

2131

D ow

nloaded from https://academ

ic.oup.com /ageing/article/50/6/2123/6358411 by 81695661, O

U P on 17 N

ovem ber 2021

C. M. Dohrendorf et al.

21. Goldenberg JZ, Yap C, Lytvyn L et al. Probiotics for the pre- vention of Clostridium difficile-associated diarrhea in adults and children. Cochrane Database Syst Rev 2017; 12: 1–207.

22. Lau CS, Chamberlain RS. Probiotics are effective at pre- venting Clostridium difficile-associated diarrhea: a systematic review and meta-analysis. Int J Gen Med 2016; 9: 27–37.

23. Mallina R, Craik J, Briffa N et al. Probiotic containing Lac- tobacillus casei, Lactobacillus bulgaricus, and Streptococcus thermophiles (ACTIMEL) for the prevention of Clostridium difficile associated diarrhoea in the elderly with proximal femur fractures. J Infect Public Health 2018; 11: 85–8.

24. Luu S, Spelman D, Woolley IJ. Post-splenectomy sepsis: preventative strategies, challenges, and solutions. Infect Drug Resist 2019; 12: 2839–51.

25. Na’amnih W, Adler A, Miller-Roll T et al. Incidence and Risk Factors for Community and Hospital Acquisition of Clostridium difficile Infection in the Tel Aviv Sourasky Medical Center. Infect Control Hosp Epidemiol 2017; 38: 912–20.

26. Freedberg DE, Salmasian H, Cohen B et al. Receipt of Antibiotics in Hospitalized Patients and Risk for Clostridium difficile Infection in Subsequent Patients Who Occupy the Same Bed. JAMA Intern Med 2016; 176: 1801–8.

27. McFarland LV, Mulligan ME, Kwok RYY et al. Nosocomial Acquisition of Clostridium-Difficile Infection. N Engl J Med 1989; 320: 204–10.

28. Djukic M, Braun LM, Unkel S et al. Introduction of Non- Vitamin K Antagonist Anticoagulants Strongly Increased the Rate of Anticoagulation in Hospitalized Geriatric Patients with Atrial Fibrillation. Drugs Aging 2018; 35: 859–69.

29. Haywood LJ, Ford CE, Crow RS et al. Atrial fibrillation at baseline and during follow-up in ALLHAT (Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial). J Am Coll Cardiol 2009; 54: 2023–31.

30. Haji-Michael P. Critical care transfers - a danger foreseen is half avoided. Crit Care 2005; 9: 343–4.

Received 9 February 2021; editorial decision 27 June 2021

2132

D ow

nloaded from https://academ

ic.oup.com /ageing/article/50/6/2123/6358411 by 81695661, O

U P on 17 N

ovem ber 2021

© 2021 British Geriatric Society. Copyright of Age & Ageing is the property of Oxford University Press / USA and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.