Season and Temperature Effects on Bloodstream Infection Incidence in a Korean Tertiary Referral Hospital

Original ArtIcle
Young Suk Sohn1Jung-Hyun Byun2Young Ah Kim3*Dong Chun Shin4Kyungwon Lee1

ABSTRACT

Background: The weather has well-documented effects on infectious disease and reports suggest that summer peaks in the incidences of gram-negative bacterial infections among hospitalized patients. We evaluated how season and temperature changes affect bloodstream infection (BSI) incidences of major pathogens to understand BSI trends with an emphasis on acquisition sites.

Methods: Incidence rates of BSIs by Staphylococcus aureus, Enterococcus spp., Escherichia coli, Klebsiella pneumoniae, Acinetobacter spp., and Pseudomonas aeruginosa were retrospectively analyzed from blood cultures during 2008–2016 at a university hospital in Seoul, Korea according to the acquisition sites. Warm months (June–September) had an average temperature of ≥20 °C and cold months (December–February) had an average temperature of ≤5 °C.

Results: We analyzed 18,047 cases, where 43% were with community-onset BSI. E. coli (N = 5,365) was the most common pathogen, followed by Enterococcus spp. (N = 3,980), S. aureus (N = 3,075), K. pneumoniae (N = 3,043), Acinetobacter spp. (N = 1,657), and P. aeruginosa (N = 927). The incidence of hospital-acquired BSI by Enterococcus spp. was weakly correlated with temperature, and the median incidence was higher during cold months. The incidence of community-onset BSI by E. coli was higher in warm months and was weakly correlated with temperature.

Conclusion: We found seasonal or temperature-associated variation in some species-associated BSIs. This could be a useful information for enhancing infection control and public health policies by taking season or climate into consideration.

Keywords



INTRODUCTION

The weather has well-documented effects on infectious disease patterns [1,2], and there is increased interest in this dynamic with growing concerns about global warming. Because climate change will also impact Korea, it is crucial to elucidate epidemiologic changes of major pathogens to better understand the association between pathogen incidence and seasonal and temperature variations. Summer peaks in the incidences of gram-negative bacterial infections among hospitalized patients have been reported [3-7]. An unusual seasonal pattern of Acinetobacter calcoaceticus was reported in the 1970s and in a previous study [5]. We also observed seasonality in community-onset Acinetobacter baumannii complex colonization or infection [8]. There is less evidence of seasonal patterns among gram-positive bacteria, but a review of laboratory blood culture data from 1961 to 1981 reported seasonal and monthly variations in Streptococcus pneumoniae epidemiology [9]. The impact of temperature change could be different according to acquisition sites, considering well-controlled cooling and heating system in healthcare facilities. In this study, we evaluated trends in the most important clinical isolates according to temperature variability.

MATERIALS AND METHODS

Blood culture analyses

We retrospectively reviewed 18,047 isolates from blood cultures that were carried out from 2008 to 2016 in a university hospital in Seoul, Korea. Incidence rates of bloodstream infections (BSIs) by Staphylococcus aureus, Enterococcus spp., Escherichia coli, Klebsiella pneumoniae, Acinetobacter spp., and Pseudomonas aeruginosa were retrospectively analyzed. We only included pathogens from BSI, such as S. aureus, Enterococcus spp., E. coli, K. pneumoniae, Acinetobacter spp., and P. aeruginosa to definitively exclude colonizers. We analyzed blood cultures from the first isolate strains among those taken from a single patient within 30-day period to eliminate duplicate results. We converted incidence rates to cases/106 patient days for hospital-acquired isolates and cases/105 patient days for community-onset isolates because we wanted to improve readability by making it less decimal. Blood culture was performed with the automated BACT/ALERT 3D system (bioMérieux, Marcy-l'Étoile, France). Species identification was performed with Bruker MS (Bruker Daltonik, Leipzig, Germany).

Definitions

Warm months (June–September) were those with an average temperature ≥20°C and cold months (December–February) were those with an average temperature ≤5°C. Monthly average temperatures (high) were obtained from the National Weather Service Forecast Office (http://www.kma.go.kr/index.jsp), and we used data, collected at the nearest observation site from the hospital location. We used monthly average temperatures over the study period, according to the previous study [8]. “Hospital-acquired” isolates were those that were obtained after 48 hours of admission, and other isolates obtained within 48 hours of admission, including outpatient isolates, were considered “community-onset (acquired)” [10].

Statistical analyses

The MedCalc Statistical Software version 17.6 (MedCalc Software bvba, Ostend, Belgium; http://www.medcalc.org; 2017) was used for all statistical analyses. Incidence differences between warm and cold months were compared using the Mann-Whitney test (independent samples). The correlations with temperature were evaluated using Pearson’s correlation analysis. In interpretation, we described the extent of correlation according to r value (if r >0.7, strong; 0.3 = < r = < 0.7, moderate; r < 0.3, week). P values <0.05 were regarded as significant.

RESULTS

Total blood culture isolates

We analyzed 18,047 cases from 2008–2016, and 43% of them were community-onset BSIs. E. coli (N = 5,365) was the most common pathogen, followed by Enterococcus spp. (N = 3,980), S. aureus (N = 3,075), K. pneumoniae (N = 3,043), Acinetobacter spp. (N = 1,657), and P. aeruginosa (N = 927).

The ratio of community-onset BSIs/hospital-acquired BSIs (CO/HA) of E. coli was 2.4. In contrast, that of S. aureus, Enterococcus spp., Acinetobacter spp., and P. aeruginosa were less than 1 suggesting that most of them were hospital-acquired BSI cases (Table 1).

Table 1. Etiologic agents of bloodstream infections according to acquisition site (2008–2016)https://www.acm.or.kr/images/acm/acm_20-004_images/Table_acm_20-004_T1.png

Abbreviation: CO, community-onset; HA, hospital-acquired.

Staphylococcus aureus and Enterococcus spp.

S. aureus caused a median 17.88 cases per 105 patient days of community-onset BSIs (interquartile range, IQR, 13.96–20.92) in warm months and 1.68 (IQR, 15.25–20.37) in cold months (P = 0.6769, Table 2). S. aureus caused a median 29.10 cases per 106 patient days of hospital-acquired BSIs (IQR, 22.50–37.10) in warm months and 26.90 (IQR, 20.80–34.70) in cold months (P = 0.6467, Table 3). Neither of the S. aureus incidence rates were correlated with temperature (community-onset BSI: correlation coefficient, r = 0.0565, P = 0.5612, 95% confidence interval, 95% CI, -0.1339 to 0.2429; hospital-acquired BSI: r = 0.0444, P = 0.6484, 95% CI, -0.1458 to 0.2314; Fig. 1).

Table 2. Incidence of community-onset bloodstream infections according to acquisition site and seasonal changes https://www.acm.or.kr/images/acm/acm_20-004_images/Table_acm_20-004_T2.png

Abbreviation: M, median; CI, confidence interval; IQR, interquartile range.

Table 3. Incidence of hospital-acquired bloodstream infections according to acquisition site and seasonal changes https://www.acm.or.kr/images/acm/acm_20-004_images/Table_acm_20-004_T3.png

Abbreviation: M, median; CI, confidence interval; IQR, interquartile range.

Enterococcus spp. caused a median 0.98 cases per 105 patient days of community-onset BSIs (IQR, 0.51–1.48) in warm months and 0.87 (IQR, 0.60–1.26) in cold months (P = 0.6871, Table 2). Enterococcus spp. caused a mean 45.20 cases per 106 patient days of hospital-acquired BSIs (IQR, 37.90–51.50) in warm months and 57.20 (IQR, 48.90–65.60) in cold months (P = 0.0021, Table 3). The incidence rate of Enterococcus spp.-caused community-onset BSI (r = 0.0952, P = 0.3273, 95% CI, -0.0955 to 0.2791) was not correlated with temperature, but the incidence rate of Enterococcus spp. of hospital-acquired BSI (r = -0.3020, P = 0.0015, 95% CI, -0.4645 to -0.1199) was weakly correlated with temperature (Fig. 1).

https://www.acm.or.kr/images/acm/acm_20-004_images/Figure_acm_20-004_F1.png

Fig. 1. Correlation between bloodstream infection (BSI) by gram-positive cocci and average monthly temperature from 2008–2016 based on Pearson’s correlation coefficient. (a) Temperature (°C) vs. incidence rate of community-onset BSI by S. aureus (cases per 105 patient days), y = 0.00337x+1.774, r = 0.0565, P = 0.5612. (b) Temperature (°C) vs. incidence rate of hospital-acquired BSI by S. aureus (cases per 106 patient days), y = 0.0588x+29.781, r = 0.0444, P = 0.6484. (c) Temperature (°C) vs. incidence rate of community-onset BSI by Enterococcus spp. (cases per 105 patient days), y = 0.0488x+9.558, r = 0.0952, P = 0.3273. (d) Temperature (°C) vs. incidence rate of hospital-acquired BSI by Enterococcus spp. (cases per 106 patient days), y = -0.495x+59.061, r = -0.3020, P = 0.0015. CA, community-onset; HA, hospital-acquired; Temp., temperature.

Escherichia coli and Klebsiella pneumoniae

E. coli caused a median of 60.40 cases per 105 patient days of community-onset BSIs (IQR, 53.10–81.30) in warm months and 52.70 (IQR, 41.10–61.50) in cold months (P = 0.0044, Table 2). E. coli caused a median of 25.40 cases per 106 patient days of hospital-acquired BSIs (IQR, 20.60–31.50) in warm months and 24.20 (IQR, 16.90–31.20) in cold months (P = 0.3241, Table 3). The incidence rate of E. coli-caused community-onset BSI (r = 0.3304, P = 0.0005, 95% CI 0.1509 to 0.4889) was weakly correlated with temperature, but the E. coli incidence rate for hospital-acquired BSI (r = 0.1443, P = 0.1363, 95% CI, -0.4596 to 0.3244) was not (Fig. 2).

K. pneumoniae caused a median of 28.20 cases per 105 patient days of community-onset BSIs (IQR, 20.90–33.70) in warm months and 22.50 (IQR, 17.00–27.90) in cold months (P = 0.0518, Table 2). K. pneumoniae caused a median of 23.50 cases per 106 patient days of hospital-acquired BSIs (IQR, 16.90–30.50) in warm months and 19.80 (IQR, 13.70–27.10) in cold months (P = 0.1965, Table 3). Neither of the K. pneumoniae incidence rates for community-onset BSI (r = 0.1986, P = 0.0394, 95% CI: 0.0099 to 0.3735) nor hospital-acquired BSI (r = 0.1094, P = 0.2596, 95% CI: -0.0812 to 0.2924) was correlated with temperature (Fig. 2).

https://www.acm.or.kr/images/acm/acm_20-004_images/Figure_acm_20-004_F2.png

Fig. 2. Correlation between bloodstream infection (BSI) by gram-negative bacilli and average monthly temperature from 2008–2016 based on Pearson’s correlation coefficient. (a) Temperature (°C) vs. incidence rate of community-onset BSI by E. coli (cases per 105 patient days), y = 0.540x+52.754, r = 0.3304, P = 0.0005. (b) Temperature (°C) vs. incidence rate of hospital-acquired BSI by E. coli (cases per 106 patient days), y = 0.128x+23.142, r = 0.1443, P = 0.1363. (c) Temperature (°C) vs. incidence rate of community-onset BSI by K. pneumoniae (cases per 105 patient days), y = 0.193x+22.874, r = 0.1986, P = 0.0394. (d) Temperature (°C) vs. incidence rate of hospital-acquired BSI by K. pneumoniae (cases per 106 patient days), y = 0.124x+20.931, r = 0.1094, P = 0.2596. CO, community-onset; HA, hospital-acquired; Temp., temperature.

Acinetobacter spp. and Pseudomonas aeruginosa

Acinetobacter spp. caused a median of 3.34 cases per 105 patient days of community-onset BSIs (IQR, 1.68–5.07) in warm months and 1.68 (IQR, 0.00–3.84) in cold months (P = 0.0672, Table 2). Acinetobacter spp. caused a median of 20.90 cases per 106 patient days of hospital-acquired BSIs (IQR, 13.80–27.70) in warm months and 26.60 (IQR, 19.70–31.40) in cold months (P = 0.0956, Table 3). Neither of the Acinetobacter spp. incidence rates for community-onset BSI (r = 0.2100, P = 0.0292, 95% CI, 0.0219 to 0.3837) nor hospital-acquired BSI (r = -0.1148, P = 0.2368, 95% CI, -0.2973 to 0.0758) was correlated with temperature (Fig. 3).

https://www.acm.or.kr/images/acm/acm_20-004_images/Figure_acm_20-004_F3.png

Fig. 3. Correlation between bloodstream infection (BSI) by glucose non-fermenters and average monthly temperature from 2008–2016. (a) Temperature (°C) vs. incidence rate of community-onset BSI by Acinetobacter spp. (cases per 105 patient days), y = 0.0500x+2.065, r = 0.2100, P = 0.0292. (b) Temperature (°C) vs. incidence rate of hospital-acquired BSI by Acinetobacter spp. (cases per 106 patient days), y = -0.121x+25.202, r = -0.1148, P = 0.2368. (c) Temperature (°C) vs. incidence rate of community-onset BSI by P. aeruginosa (cases per 105 patient days), y = 0.000520x+0.455, r = 0.0137, P = 0.8879. (d) Temperature (°C) vs. incidence rate of hospital-acquired BSI by P. aeruginosa (cases per 106 patient days), y = 0.0565x+8.814, r = -0.0957, P = 0.3241. CO, community-onset; HA, hospital-acquired; Temp., temperature.

P. aeruginosa caused a median of 0.35 cases per 105 patient days of community-onset BSIs (IQR, 0.17–0.85) in warm months and 0.40 (IQR, 0.31–0.78) in cold months (P = 0.5976, Table 2). P. aeruginosa caused a median of 8.87 cases per 106 patient days of hospital-acquired BSIs (IQR, 5.57–12.7) in warm months and 7.97 (IQR, 5.13–12.31) in cold months (P = 0.3241, Table 3). Neither of the P. aeruginosa incidence rates for community-onset BSI (r = 0.0137, P = 0.8879, 95% CI, -0.1757 to 0.2022) nor hospital-acquired BSI (r = -0.0957, P = 0.3241, 95% CI, -0.0949 to 0.2797) was correlated with temperature (Fig. 3).

DISCUSSION

Summer peaks in the incidences of gram-negative bacterial infections relative to gram-positive infections among hospitalized patients have been reported [3-7]. We found no correlation between temperature and the incidence of community-onset S. aureus BSI, hospital-acquired S. aureus, or community-onset Enterococcus spp. BSI. However, the incidence of Enterococcus spp. of hospital-acquired BSI was weakly correlated with temperature, and the median value of Enterococcus spp.-caused hospital-acquired BSIs was higher in cold months. These results are particularly interesting, considering that the temperatures of healthcare facilities are relatively well controlled. Further studies are needed to evaluate other risk factors of hospital-acquired Enterococcus spp. BSIs.

The widespread presence of ESBL-producing E. coli in the community is well established due to the worldwide increase of the sequence type (ST) 131 clone in the mid 2000s [11]. We also found a high prevalence of ESBLs mainly CTX-M among the community-onset E. coli isolates from Korean community hospitals [12]. Recently, community-onset E. coli infections and asymptomatic carriage in healthy individuals without recent exposure to healthcare facilities have increased [13]. Multiple sources of community-based multiple drug-resistant E. coli have been suspected recently, including among livestock, companion animals, sewage, wastewater, and recreational waterways [14]. We observed a higher incidence of community-onset E. coli BSI in warm months, but this trend was not observed with the hospital-acquired E. coli BSIs. E. coli was the most common causative agent of community-onset BSIs. We think that it is a possible reason why community-onset E. coli BSIs showed seasonality in this study. Microorganisms in the community or environment could be more easily affected by temperature than those in healthcare facilities. Therefore, it is easier to observe seasonality in isolates from community, considering that the temperature varies widely than that of the healthcare facilities.

Regarding glucose non-fermenters, there is some evidence of seasonality in A. baumannii infections [3,4,8]. We found that the median number of BSIs by Acinetobacter spp. was not different between warm months and cold months, regardless of hospital-acquired or community-onset infection acquired site. Furthermore, the Acinetobacter spp. BSI incidence rate was not correlated with temperature. This result is inconsistent with our previous study, which showed a seasonal pattern for community-onset A. baumannii complex colonization or infection [8]. The discrepancy may be due to the different study designs we employed because the previous study included all infection types, while only BSIs were included in this study. Although extra-hospital reservoirs of A. baumannii complex have been suspected [15], A. baumannii primarily causes healthcare-associated infection. A. baumannii isolates dramatically increased in Korean hospitals [16]. Hospital-acquired A. baumannii complex isolation is highly clonal, indicating that the incidence of healthcare-onset cases may have been affected by the endemic nature of Korean hospitals. The median number of P. aeruginosa BSIs was not different between warm and cold months, regardless of the infection site. The P. aeruginosa BSI incidence rate was not correlated with temperature.

Although seasonal patterns of gram-negative organisms have been observed, the mechanisms underlying seasonal variation are not well understood. A possible explanation is that higher temperatures may facilitate increased bacterial growth in the environment as well as enhanced virulence of gram-negative bacteria, thereby contributing to increases in infection incidence in warmer periods [17]. Another explanation is that the lipidA moiety of lipopolysaccharides, which forms the outer monolayer of the outer membrane of most gram-negative bacteria, is regulated by environmental conditions [18]. Recently, it was reported that the carbapenemase gene, KPC showed the highest frequency of gene transfer at 25°C and NDM at 30°C, which suggested temperature was related to plasmid transfer frequency [19].

There are some limitations as a single-center study, and thus, further study is required in diverse clinical settings. However, we found that E. coli (N = 5,365) was the most commonly isolated from BSI, followed by Enterococcus spp. (N = 3,980), S. aureus (N = 3,075), K. pneumoniae (N = 3,043), Acinetobacter spp. (N = 1,657), and P. aeruginosa (N = 927), which was consistent with previous studies [20-22]. Based on this data, we think that bias was unlikely as a single agency study.

In conclusion, we found seasonal or temperature-associated variation in BSIs according to some species. Our findings could be useful information for enhancing infection control and public health policies that take season or climate into consideration. The enhancement of infection control and public health policies, taking season or climate into consideration could be needed.

요약

배경: 감염질환의 빈도가 날씨에 따라 달라지며, 입원환자에서 여름철에 그람음성 세균 감염이 증가한다는 것이 알려져 있다. 본 연구에서는 주요 병원균에 의한 혈류감염의 빈도가 계절 및 온도에 따라 달라지는지, 지역사회 감염과 병원감염을 나누어 평가하고자 하였다.

방법: 2008~2016년 서울의 한 대학병원에서 발생한 Staphylococcus aureus, Enterococcus spp., Escherichia coli, Klebsiella pneumoniae, Acinetobacter spp. 및 Pseudomonas aeruginosa에 의한 혈류감염을 후향적으로 분석하였다. 여름(6~9월)은 평균 온도가 20°C 이상으로, 겨울 (12~2월)은 평균 온도가 5°C 미만으로 정의하였다.

결과: 지역사회 발생 E. coli 혈류감염이 총 18,047건 중 43% (N = 5,365)를 차지하여 가장 흔했고, Enteroccus spp. (N = 3,980), S. aureus (N = 3,075), K. pneumoniae (N = 3,043), Acinetobacter spp. (N = 1,657), 및 P. aeruginosa에 의한 혈류감염 (N = 927) 순이었다. 병원에서 발생한 Enteroccus spp. 혈류감염은 온도와 약한 상관관계를 보였고, 추운 달에 발생률이 더 높았다. 지역사회 발생 E. coli 혈류감염은 여름에 발생률이 더 높았고, 온도와 약하게 상관관계가 있었다.

결론: 일부 균종에서 혈류감염의 발생이 계절적 또는 온도와 연관됨을 알 수 있었다. 이는 계절이나 기후를 고려한 감염 관리 및 공중 보건 정책을 강화하는 데 유용한 정보가 될 수 있다.

ACKNOWLEDGEMENTS

The authors were grateful to Do Yong Kwak for electronic data collection.

REFERENCES

1  1. Fisman DN. Seasonality of infectious diseases. Annu Rev Public Health 2007;28:127-43.  

2  2. Polgreen PM, Polgreen EL. Infectious diseases, weather, and climate. Clin Infect Dis 2018;66:815-7.  

3  3. Eber MR, Shardell M, Schweizer ML, Laxminarayan R, Perencevich EN. Seasonal and temperature-associated increases in gram-negative bacterial bloodstream infections among hospitalized patients. PLoS One 2011;6:e25298  

4  4. Perencevich EN, McGregor JC, Shardell M, Furuno JP, Harris AD, Morris JG, et al. Summer peaks in the incidences of gram-negative bacterial infection among hospitalized patients. Infect Control Hosp Epidemiol 2008;29:1124-31.  

5  5. Retailliau HF, Hightower AW, Dixon RE, Allen JR.Acinetobacter calcoaceticus : a nosocomial pathogen with an unusual seasonal pattern. J Infect Dis 1979;139:371-5.  

6  6. Richet H. Seasonality in gram-negative and healthcare-associated infections. Clin Microbiol Infect 2012;18:934-40.  

7  7. Rodrigues FS, Clemente de Luca FA, Ribeiro da Cunha A, Fortaleza CMCB. Seasons, weather and predictors of healthcare-associated gram-negative bloodstream infections: a case-only study. J Hosp Infect 2019;101:134-41.  

8  8. Kim YA, Kim JJ, Won DJ, Lee K. Seasonal and temperature-associated increase in community-onset Acinetobacter baumannii complex colonization or infection. Ann Lab Med 2018;38:266-70.  

9  9. Flournoy DJ, Stalling FH, Catron TL. Seasonal and monthly variation of Streptococcus pneumoniae and other pathogens in bacteremia (1961-1981). Ecol Dis 1983;2:157-60.  

10  10. Friedman ND, Kaye KS, Stout JE, McGarry SA, Trivette SL, Briggs JP, et al. Healthcare-associated bloodstream infections in adults: a reason to change the accepted definition of community-acquired infections. Ann Intern Med 2002;137:791-7.  

11  11. Colpan A, Johnston B, Porter S, Clabots C, Anway R, Thao L, et al.Escherichia coli Sequence type 131 (ST131) subclone H30 as an emergent multidrug-resistant pathogen among US veterans. Clin Infect Dis 2013;57:1256-65.  

12  12. Kim YA, Kim JJ, Kim H, Lee K. Community-onset extended-spectrum-β-lactamase-producing Escherichia coli sequence type 131 at two Korean community hospitals: the spread of multidrug-resistant E. coli to the community via healthcare facilities. Int J Infect Dis 2017;54:39-42.  

13  13. Spellberg B, Doi Y. Editor's choice: the rise of fluoroquinolone-resistant Escherichia coli in the community: scarier than we thought. J Infect Dis 2015;212:1853-5.  

14  14. Lazarus B, Paterson DL, Mollinger JL, Rogers BA. Do human extraintestinal Escherichia coli infections resistant to expanded-spectrum cephalosporins originate from food-producing animals? a systematic review. Clin Infect Dis 2014;60:439-52.  

15  15. Eveillard M, Kempf M, Belmonte O, Pailhoriès H, Joly-Guillou ML. Reservoirs of Acinetobacter baumannii outside the hospital and potential involvement in emerging human community-acquired infections. Int J Infect Dis 2013;17:e802-5.  

16  16. Lee Y, Kim YA, Song W, Lee H, Lee HS, Jang SJ, et al. Recent trends in antimicrobial resistance in intensive care units in Korea. Korean J Nosocomial Infect Control 2014;19:29-36.  

17  17. Ratkowsky D, Lowry R, McMeekin T, Stokes A, Chandler R. Model for bacterial culture growth rate throughout the entire biokinetic temperature range. J Bacteriol 1983;154:1222-6.  

18  18. Raetz CR, Reynolds CM, Trent MS, Bishop RE. Lipid a modification systems in gram-negative bacteria. Annu Rew Bochem 2007;76:295-329.  

19  19. Yang J, Bae S, Park C, Lee K. Presented at the Korean society of infectious diseases and Korean society of antimicrobial therapy, 2018  

20  20. Ahn GY, Lee SH, Jeong OY, Chaulagain BP, Moon DS, Park YJ. Trends of the species and antimicrobial susceptibility of microorganisms isolated from blood cultures of patients. Korean J Clin Microbiol 2006;9:42-50.  

21  21. Kim N, Hwang J, Song K, Choe PG, Park WB, Kim ES, et al. Changes in antimicrobial susceptibility of blood isolates in a university hospital in South Korea, 1998-2010. Infect Chemother 2012;44:275-81.  

22  22. Kim SY, Lim G, Kim MJ, Suh JT, Lee HJ. Trends in five-year blood cultures of patients at a university hospital (2003~ 2007). Korean J Clin Microbiol 2009;12:163-8.