Epidemiology

There’s an age-old problem in science: how do you prove a causal relationship between variables that correlate? Proving that the variables are correlated is the easy part; it’s more difficult to disentangle cause from effect. This can be seen in several studies that identify a correlation between environmental burden and the number of patients that are infected or colonized with pathogens.Figure 1. Correlation between the number of patients infected with Acinetobacter spp. and the number of positive Acinetobacter spp. environmental cultures per calendar month during an outbreak on a neurosurgical ICU.¹

Figure 2. Correlation between microbial burden and the number of patients who acquired an HAI in ICUs.²

Figure 3. Correlation between the number of hygiene failures and the number of patients who acquired an infection on a surgical intensive care unit each week.³

So can we conclude that the higher burden of contamination resulted in an increased risk of acquisition? Or is it that more patients were infected or colonized with pathogens, which resulted in more environmental shedding? From these studies, you can’t be sure.

If you were seeking to prove the role of a gene in a process, you’d knock out the gene and demonstrate that the process stopped or changed. So, the only way to disentangle cause and effect in contamination and transmission is to perform an intervention to reduce environmental contamination and show that this correlates with reduced transmission. While the Salgado study evaluated an intervention, the data correlating contamination burden with HAIs was not stratified by the intervention, which would have been one way to assess likely causation.²

There is some further in vitro and epidemiological data supporting that the degree of transmission may be proportional to the environmental burden. An in vitro mouse model established a ‘dose-response’ relationship between the degree of contamination with C. difficile spores and the development of CDI.⁴ Furthermore, this model showed that disinfectants that achieved a greater log reduction of C. difficile spores were more able to interrupt transmission.

Also, one of the studies demonstrating that admission to a room previously occupied by a patient with VRE increases the chances of VRE acquisition identified something amounting to a ‘dose response’.⁵ The greatest increased risk was for patients admitted to a room with an environmental culture positive for VRE, and being admitted to a room where the immediate prior room occupant was colonized with VRE carried a greater increased risk than being admitted to a room where any patient in the 2 weeks prior to admission was VRE colonized (Figure 4).

Figure 4. How the increased risk of acquiring VRE from the prior room occupant changes due to patient and environmental factors.⁵

Is there a causal relationship between contamination burden and transmission risk? On balance, the answer seems to be yes, though it would be useful to have a solid intervention study to prove that an increasing environmental burden causes an incrementally increase in transmission risk.

Article citations:

Denton M, Wilcox MH, Parnell P et al. Role of environmental cleaning in controlling an outbreak of Acinetobacter baumannii on a neurosurgical intensive care unit. J Hosp Infect 2004; 56: 106-110.
Salgado CD, Sepkowitz KA, John JF et al. Copper surfaces reduce the rate of healthcare-acquired infections in the intensive care unit. Infect Control Hosp Epidemiol 2013; 34: 479-486.
White LF, Dancer SJ, Robertson C, McDonald J. Are hygiene standards useful in assessing infection risk? Am J Infect Control 2008; 36: 381-384.
Lawley TD, Clare S, Deakin LJ et al. Use of purified Clostridium difficile spores to facilitate evaluation of health care disinfection regimens. Appl Environ Microbiol 2010; 76: 6895-6900.
Drees M, Snydman D, Schmid C et al. Prior environmental contamination increases the risk of acquisition of vancomycin-resistant enterococci. Clin Infect Dis 2008; 46: 678-685.

I am no expert in HIV, but I know that ‘vertical transmission’ means something very specific:

Vertical transmission: the transmission of a disease from mother to child either during pregnancy, childbirth, or by breastfeeding.

Similarly, the definition of ‘horizontal transmission’ is well defined:

Horizontal transmission: the transfer of an infection from person to person.

So, when I read about ‘vertical’ and ‘horizontal’ interventions in a recent New England Journal of Medicine Editorial and the Controversies blog, I began to get a little confused. I have a PhD in epidemiology so don’t consider myself easy to confuse (in this particular domain), but I would have thought that a ‘horizontal intervention’ would be directed towards preventing horizontal spread of an infectious agent and a ‘vertical intervention’ would be directed towards preventing the vertical transmission of an infectious agent. But this is not how these terms are being applied. Instead, a ‘horizontal intervention’ is being used to describe an intervention applied to every patient (such as chlorhexidine bathing or hospital-wide hand hygiene interventions) whereas a ‘vertical intervention’ is being used to describe an intervention designed to reduce colonization or infection due to a specific pathogen (such as active screening and isolation to prevent the spread of MRSA). The use of the term ‘vertical intervention’ seems especially confusing, since it’s a ‘vertical intervention’ to prevent the horizontal transmission of a specific pathogen!

I fail to see how the terms ‘vertical’ or ‘horizontal’ intervention are useful when there are such well-established definitions for horizontal and vertical transmission. I think that ‘universal intervention’ (such as universal screening or decolonization) and ‘targeted intervention’ (such as active screening and isolation to prevent the spread of MRSA) make a lot more sense. These terms are already in common circulation, so I would urge those who favour the use of ‘vertical’ or ‘horizontal’ intervention to reconsider their terminology.