Role of Rigors in Bacteremia – Do Not Ignore Chills

Blood cultures are typically drawn in patients who present with sepsis due to concern for infection. While the trigger response to “fevers and leukocytosis” is to obtain blood cultures, the positive rate that has been quoted ranges from 4-7% (1), depending on the source. Indeed, blood culture contamination is a frequent problem that is encountered anywhere from 20-56% of cases where blood cultures are obtained (2). As a result, multiple algorithms and predictive tools have been developed to ensure blood cultures are appropriately ordered in those who have a high pretest probability of bacteremia. 

One of the simplest predictive tools I’ve found are rigors. These are muscle  contractions that result in heat production, which may play a role in aiding fever onset (3). Given that fever following exposure to endotoxin can be delayed anywhere from 3-5 hours (4) and chills/rigors can be seen prior to fever (5), there may be a correlation between bacteremia and the presentation of rigors. The association may be much stronger than that between fever and bacteremia. This has been investigated in several cohort studies. 

An early cohort study evaluated several variables to generate a predictive score for bacteremia in patients presenting with sepsis (6). Using a dataset of 1342 episodes of sepsis, chills occurred more often in patients with true positive blood culture results (27%) compared to all other results (19%, p-value 0.011). Logistic regression analysis found that chills had an OR of 1.7 (95% CI 1.0-2.8) for gram negative rod bacteremia, though this did not hold for any bacteremia or gram positive bacteremia. Another algorithm (7) developed for bacteremia in patients with acute febrile illness found that chills occurred more often in patients with bacteremia compared to those without bacteremia (75% vs 27.8%, p-value <0.001). Univariate analysis found that chills, along with old age, high body temperature and high CRP were associated with bacteremia. The authors used recursive partitioning analysis to generate 2 algorithms, one involving all clinical signs/symptoms while the second used laboratory analysis, including CRP and leukocytosis. In both clinical scenarios, the relative importance score for chills were 100. Indeed, both algorithms required chills:

And all patients with high risk, who had chills, were more likely to have bacteremia, usually around a quarter of the time:

This highlights the predictive power of chills/rigors in context of fevers. Other studies have looked at the likelihood of positive blood cultures when patients present with rigors. For instance, a prospective study (8) evaluated 396 febrile adults who presented to the ED. Multivariate analysis found that rigors had an OR of 13.7 (95% CI 4.47-42.0) for bacteremia, compared to a temperature >39.9C, which had an OR of 2.68 (95% CI 1.03-6.94). 

Another single center prospective study (9) enrolled patients presenting with fevers less than 2 weeks duration, and found that, while higher temperatures (higher than 38.8C) were associated with higher risk of bacteremia, moderate chills (feeling very cold, needing a thick blanket) had a RR of 4.14 (95% CI 1.6-10.656) while shaking chills (defined as feeling extremely cold with rigors and generalized body shaking under a thick blanket) had an adjusted RR of 12.12 (95% CI 4.1-36.2). 

Increasing severity of rigors also increased the positive likelihood ratio, with lack of chills (compared to mild chills) having a modest negative likelihood ratio of 0.24:

A multicenter cohort study also confirmed this association (10). This was a sub analysis of the HERO study (a prospective multicenter study evaluating the validity of several inflammatory markers in predicting organ dysfunction) which evaluated the utility of shaking chills and vomiting in predicting bacteremia. Logistic regression was used to generate 3 models, with model 1 being a univariate analysis, model 2 adjusting for sex and age, and model 3 adjusting for the use of prior antibiotics. In all 3 models, chills were associated with higher likelihood of bacteremia:

Another logistic regression was fitted to account for viral infections (models 4 and 5), which resulted in a higher odds ratio for bacteremia:

A more useful way to apply this data is based on PLR and NLR. The presence of chills yields a LR of 2.47, however their absence does not completely rule out the presence of bacteremia. 

The sooner after the rigors occur you get the blood cultures, the higher the yield. A prospective cohort study (11) found that obtaining blood cultures within 2 hours after rigors yielded a higher rate of blood culture positivity when compared to those obtained after 2 hours (54% vs 38%, p-value 0.019). This remained statistically significant after adjusting for antibiotic exposure (56% vs 40%, p-value 0.030). OR was higher for those who had more than one episode of shaking chills and if blood cultures were obtained within 2 hours:

A large, well known review of 35 studies (12) found that fevers were not terribly helpful to predict bacteremia.In contrast, chills are more useful with overall positive LR of 2.2 (95% CI 1.4-3.3) compared to subjective fevers. Notably, the more severe teh chills, the higher the positive LR. For fevers, the positive LR was not terribly helpful when compared to overall chills:

That fevers by themselves not being good predictors for bacteremia was also seen in a single-center VA study of inpatients who had a blood culture obtained (13). When looking at likelihood ratios, fever as an indication did not predict blood culture positivity, even with additional indications or leukocytosis (though lack of antibiotic exposure and fever improved the likelihood ratio):

At the end of the day, paying attention to chills while using this with the overall clinical picture may be way more useful to guide the use of blood culture in contrast to fevers alone. You may want to query about rigors or shaking chills rather than fever, as this may be a bigger clue for possible bacteremia. 

References:

1. Coburn B, Morris AM, Tomlinson G, Detsky AS. Does This Adult Patient With Suspected Bacteremia Require Blood Cultures? JAMA. 2012;308(5):502–511. doi:10.1001/jama.2012.8262
2. Doern GV, Carroll KC, Diekema DJ, Garey KW, Rupp ME, Weinstein MP, Sexton DJ. Practical Guidance for Clinical Microbiology Laboratories: A Comprehensive Update on the Problem of Blood Culture Contamination and a Discussion of Methods for Addressing the Problem. Clin Microbiol Rev. 2019 Oct 30;33(1):e00009-19. doi: 10.1128/CMR.00009-19. PMID: 31666280; PMCID: PMC6822992.
3. van Ooijen AM, van Marken Lichtenbelt WD, van Steenhoven AA, Westerterp KR. Cold-induced heat production preceding shivering. Br J Nutr. 2005 Mar;93(3):387-91. doi: 10.1079/bjn20041362. PMID: 15877879.
4. Greisman SE, Hornick RB. Comparative pyrogenic reactivity of rabbit and man to bacterial endotoxin. Proc Soc Exp Biol Med. 1969 Sep;131(4):1154-8. doi: 10.3181/00379727-131-34059. PMID: 4897836.
5. Guieu JD, Hellon RF. The chill sensation in fever. Pflugers Arch. 1980 Mar;384(1):103-4. doi: 10.1007/BF00589522. PMID: 7189863.
6. Bates DW, Sands K, Miller E, Lanken PN, Hibberd PL, Graman PS, Schwartz JS, Kahn K, Snydman DR, Parsonnet J, Moore R, Black E, Johnson BL, Jha A, Platt R. Predicting bacteremia in patients with sepsis syndrome. Academic Medical Center Consortium Sepsis Project Working Group. J Infect Dis. 1997 Dec;176(6):1538-51. doi: 10.1086/514153. PMID: 9395366.
7. Tokuda Y, Miyasato H, Stein GH. A simple prediction algorithm for bacteraemia in patients with acute febrile illness. QJM. 2005 Nov;98(11):813-20. doi: 10.1093/qjmed/hci120. Epub 2005 Sep 20. PMID: 16174688.
8. Lee CC, Wu CJ, Chi CH, Lee NY, Chen PL, Lee HC, Chang CM, Ko NY, Ko WC. Prediction of community-onset bacteremia among febrile adults visiting an emergency department: rigor matters. Diagn Microbiol Infect Dis. 2012 Jun;73(2):168-73. doi: 10.1016/j.diagmicrobio.2012.02.009. Epub 2012 Mar 29. PMID: 22463870.
9. Tokuda Y, Miyasato H, Stein GH, Kishaba T. The degree of chills for risk of bacteremia in acute febrile illness. Am J Med. 2005 Dec;118(12):1417. doi: 10.1016/j.amjmed.2005.06.043. PMID: 16378800.
10. Holmqvist M, Inghammar M, Påhlman LI, Boyd J, Åkesson P, Linder A, Kahn F. Risk of bacteremia in patients presenting with shaking chills and vomiting – a prospective cohort study. Epidemiol Infect. 2020 Mar 31;148:e86. doi: 10.1017/S0950268820000746. PMID: 32228723; PMCID: PMC7189349.
11. Taniguchi T, Tsuha S, Shiiki S, Narita M. High positivity of blood cultures obtained within two hours after shaking chills. Int J Infect Dis. 2018 Nov;76:23-28. doi: 10.1016/j.ijid.2018.07.020. Epub 2018 Jul 27. PMID: 30059771.
12. Coburn B, Morris AM, Tomlinson G, Detsky AS. Does This Adult Patient With Suspected Bacteremia Require Blood Cultures? JAMA. 2012;308(5):502–511. doi:10.1001/jama.2012.8262
13. Linsenmeyer K, Gupta K, Strymish JM, Dhanani M, Brecher SM, Breu AC. Culture if spikes? Indications and yield of blood cultures in hospitalized medical patients. J Hosp Med. 2016 May;11(5):336-40. doi: 10.1002/jhm.2541. Epub 2016 Jan 13. PMID: 26762577.