Costs of hospital-acquired Clostridium difficile infections: an analysis on the effect of time-dependent exposures using routine and surveillance data

Table 1 Descriptive statistics

	1		2		3		4
	1		2		Included in analysis with time-to-exposure restriction
	All		Patient with CDI relevant main diagnosis		Controls		Cases with nosocomial CDI
	Mean/N	SD	Mean/N	SD	Mean/N	SD	Mean/N	SD
Number of different main diagnosis groups	5878		336		296		296
Costs (real)	5154	10,895	8278	15,235.38	24,407	36,589	34,749	44,208
Reimbursement (real)	5188	10,479	8198	14,390.32	22,641	34,479	33,515	48,453
Reimbursement-costs (real)	33	3646	− 79	4858.83	− 1765	10,584.12	− 1233	13,145
LOS	7.6	8.86	10.7	11.53	25.7	21.05	34.7	28.1
Days in hospital at the time of CDI acquisition							16.99	16.39
Age	57.70	18.68	62.70	16.46	61.04	16.48	63.61	15.76
CCI	1.90	2.80	3.48	3.32	4.37	3.54	4.83	3.36
Died	2.02%		5.05%		7.43%		11.98%
Intensive care (h)	104.8	340.8	158.5	439.5	456.8	948.1	806	1352
Comorbidities	N/%		N/%		N/%		N/%
Number of SD (N)	6.58		9.27		14.91		20.53
Renal failure (N.18)	7.3%		12.4%		17%		19.6%
Heart failure (I.50)	2.7%		4.2%		6%		8.2%
Ischemic heart disease (I.25)	10.1%		14.5%		14%		17.2%
Diabetes (E.11)	10.9%		15.3%		15%		18.5%
Hypertension (I.10)	31.7%		37.2%		34%		38.5%
Atrial fibrillation and flutter (I.48)	7.8%		12.6%		15%		20.5%
Anaemia (D.62)	5.9%		9.8%		20%		27.5%
COPD (J.44)	4.1%		6.5%		7%		8.1%
Cardiac/vascular implants, grafts (Z.95)	8.7%		13.4%		15%		14.7%
Cancer (C.)	7.4%		10.2%		12%		14.3%
Observations	204,914		51,857		1951		559

Column 1 gives the descriptive statistics for all available data. Column 2 restricts it to patients in main diagnosis groups where there is at least 1 case of CDI, all diagnosis groups in which there is no case are excluded. Column 3 shows the controls used in the regression model 2 in Table 2 below. There are 40 main diagnosis groups in which there were no suitable controls for CDI cases when applying the time-to-exposure restriction. Column 4 presents all patients with CDI that are considered in regression model 2 in Table 2. Here, CDI cases which have no suitable controls that have stayed at least as long as the time of CDI exposure are deleted. There are 3 cases of CDI for which no appropriate controls could be found. These cases were therefore excluded from the analyses shown in regression model 2 in Table 2. For 93 out of the 559 cases there were less than 4 controls available for matching. Comorbidities show the 10 comorbidities as defined in Resch (2008). CCI gives the Charlson Comorbidity Score

ISSN: 1478-7547