Jpn. J. Infect. Dis., 52, 1999

Laboratory and Epidemiology Communications

More Beds: More Nosocomial Infections

Yoshimitsu Yanaka, Naho Tusaka and Kenji Yamamoto*

Laboratory of Clinical Microbiology and Immunology,
Bun'in Hospital of University of Tokyo, Mejirodai 3-28-6, Bunkyo-ku, Tokyo

Communicated by Hiroshi Yoshikura

(Accepted October 6, 1999)

Nosocomial infection is sometimes fatal. About 15% of hospital death could be related to pneumonia caused by nosocomial infections (1) which prolong patients' hospital stays and increase medical expenditure directly or indirectly. According to a study conoducted in the United States, nosocomial infection accounted for 15% of all hospital charges (1). Jarvis et al. (2,3) reported that the incidence of nosocomial infection was higher in hospitals with higher numbers of beds. They speculated that the differential risk was a by-product of severity of illness and the frequency of invasive diagnostic and therapeutic modalities. We recently observed a similar phenomenon in two teaching hospitals, referred to here as hospital A and hospital B, affiliated with a university. They were located in a city's central district and provided similar kinds of clinical services, such as internal medicine, surgery, pediatrics, gyocology, obstetrics, dermatology, nasopharyngology, orthopedics, ophthalmology, intensive care unit, etc. Severity of illness of patients, contents of medical interventions, etc. were almost identical in the two hospitals. The average patients stay was nearly the same. Area per bed (including public space for inpatients) was larger in hospital A (Table 1). The pattern of antibiotic resistance which reportedly reflects antibiotic use (4) was identical in the two hospitals (Fig. 1). Hospital A had more beds; 984 beds in hospital A vs. 245 beds in hospital B.

Methicillin resistant Staphylococcus aureus (MRSA) is one of the agents causing serious nosocomial infections. MRSA incidence, both incidence per bed and incidence per discharge (Table 1), was about three-fold higher in hospital A. Isolation of methicillin sensitive Staphylococcus aureus (MSSA) per discharge was lower for hospital A, while the ratio of MRSA to MSSA was more than three-fold higher in hospital A. All these indicators pointed to a heavier MRSA infestation in hospital A. Why was the larger hospital more prone to MRSA epidemic? The sole difference between the two hospitals appeared to be the number of beds. If the number of beds is significant, a proper mathematical model should explain what we observed here.

Our model is based on the following assumptions: 1) All the beds (b) are occupied, and there is no new admission nor discharge during the study. The number of MRSA carriers at time 0 is m0. 2) Infection by MRSA is a random process. The number of individuals infected by MRSA emitted by a carrier follows a Poisson distribution with average hit ă (constant). From time t to t+1, there will be M(b,t) such events, where M(b,t) is the number of MRSA carriers at time t. 3) The number of MRSA carriers at time t+1 is expressed by the formula M(b, t+1) = (1-k)M(b,t) + R(b,t), where R(b,t) is the increase of MRSA infected individuals during time t to time t+1 and k is the MRSA cure rate during the same period. The simulation program was written in C Language on a Sun Ultra Enterprise 450 with SunOS 5.6 (Sun Micro Systems, Palo Alto, Calif.).


Figure 2A shows 50 simulations for a hospital with 100 beds and 10 MRSA carriers at time 0 (b=100, m0=10). Here, ă (average hit) in the Poisson distribution was 1.05, and k was 0.95. MRSA outbreak occurred in 4 out of 50 trials. Figure 2B shows simulations for hospitals with 100, 250, 500, and 1,000 beds with the same ă and k values and with the same number of MRSA carriers (m0=10) at the start. Here, the probability of nosocomial infection outbreak (outbreak is defined by MRSA infection in 40% or more of the patients) is plotted against the number of beds. The probability of the outbreak increases with the number of beds. Therefore, even if two hospitals different in size take the same hospital infection control measures, i.e., k and ă values being the same, the hospital with more beds has a higher probability of experiencing an outbreak. The integration of smaller hospitals into a larger hospital was once considered economically advantageous, and such a policy has been implemented in Japan since the early 1980s. Our analysis, however, warns against such a policy. The simulation program is available from our laboratory.

REFERENCES

  1. Inlamder, C.B., Levin, L.S. and Weiner, E.D. (1998): Medicine on Trial. p.123-134. Pantheon Books, N.Y.
  2. Hughes, J.M., Culver, D.H., White, J.W., Jarvis, W.R., Morgan, W.M., Munn, V.P. and Emori, T.G.: (1983): Nosocomial infection surveillance, 1980-1982. Morbid. Mortal. Wkly. Rep., 32 (4SS), 1SS-16SS.
  3. Jarvis, W.R., White, J.W., Munn, V.P., Emori, T.G., Culver, D.H., Thornsberry, C. and Hughes, J.M. (1984): Nosocomial infection surveillance, 1983. Morbid. Mortal. Wkly. Rep., 33 (2SS), 9SS-21SS.
  4. Kizu, J., Arano, K., Miwa, H., Yamamoto, K. and Kawana, S. (1995). MRSA carriers and use of antibotics. Nihon-Kankyo-Kansen (Jpn. J. Environ. Infect.), 10, 15-22 (in Japanese).

*Corresponding author: Fax:+81-3-3816-4802, E-mail:backen@camille.is.s.u-tokyo.ac.jp

Go to JJID Homepage                                Go to JJID 52(4)