Evaluation of Hospital Nurse-to-Patient Staffing Ratios and Sepsis Bundles on Patient Outcomes

Abstract

Background.

Despite nurses’ responsibilities in recognition and treatment of sepsis, little evidence documents whether patient-to-nurse staffing ratios are associated with clinical outcomes for patients with sepsis.

Methods.

Using linked data sources from 2017 including MEDPAR patient claims, Hospital Compare, American Hospital Association, and a large survey of nurses, we estimate the effect of hospital patient-to-nurse staffing ratios and adherence to the Early Management Bundle for patients with Severe Sepsis/Septic Shock SEP-1 sepsis bundles on patients’ odds of in-hospital and 60-day mortality, readmission, and length of stay. Logistic regression is used to estimate mortality and readmission, while zero-truncated negative binomial models are used for length of stay.

Results.

Each additional patient per nurse is associated with 12% higher odds of in-hospital mortality, 7% higher odds of 60-day mortality, 7% higher odds of 60-day readmission, and longer lengths of stay, even after accounting for patient and hospital covariates including hospital adherence to SEP-1 bundles. Adherence to SEP-1 bundles is associated with lower in-hospital mortality and shorter lengths of stay; however, the effects are markedly smaller than those observed for staffing.

Discussion.

Improving hospital nurse staffing over and above implementing sepsis bundles holds promise for significant improvements in sepsis patient outcomes.

BACKGROUND

For nearly a decade, New York state has been a leader in efforts to reduce high rates of mortality for people with sepsis. In 2013, after the highly publicized death of a 12-year-old boy with sepsis, New York enacted Rory’s Regulations,¹ which required hospitals to implement evidence-based protocols for the screening, early diagnosis, and timely treatment of patients with severe sepsis and/or septic shock. Included in Rory’s Regulations was the public reporting of hospital adherence to protocols. More recently, hospitals in all states followed suit when the Centers for Medicare and Medicaid Services (CMS) implemented a similar public reporting measure based on evidence-based guidelines from the Surviving Sepsis Campaign.^2,3 Today, Hospital Compare publicly reports hospitals’ adherence scores on the Early Management Bundle for patients with Severe Sepsis/Septic Shock (SEP-1).² Despite the national adoption of evidence-based protocols for the care of patients with sepsis, few hospitals are consistently delivering the requisite care for sepsis patients,^4,5 leading to potentially preventable deaths.^6–8

Another evidence-based intervention that has received little attention in the context of caring for patients with sepsis but has been associated with better clinical outcomes for patients with various medical and surgical conditions is patient-to-nurse staffing ratios.^9–11 Some previous research has shown nurse staffing to be associated with the incidence of hospital acquired infections.¹² Less is known about the associations between patient-to-nurse staffing ratios and clinical outcomes for patients with sepsis; however, some recent research suggests that sepsis patients admitted to hospitals with better nursing resources, including better staffing ratios, have better clinical outcomes including lower odds of mortality, readmission, intensive care unit utilization, shorter lengths of stay, and lower costs of care.¹³ No research prior to this study has considered the association of patient-to-nurse staffing ratios and recommended evidence-based sepsis care bundles on outcomes for sepsis patients.

In this study, we directly evaluate whether patient-to-nurse staffing ratios are associated with clinical outcomes for patients admitted with sepsis in 116 New York state hospitals. We simultaneously evaluate the effects of hospital adherence to the SEP-1 evidence-based care bundle on patient outcomes to determine whether and to what extent improving patient-to-nurse staffing ratios might benefit patients. This research question is timely and policy relevant since New York state requires sepsis bundles and is currently considering the Safe Staffing for Quality Care Act (A2954/S1032),¹⁴ which would require hospitals to comply with safe nurse staffing ratios.

MATERIALS AND METHODS

Design and Data Sources

A cross-sectional analysis of multiple linked data sources was undertaken. Data about hospitals were provided from several sources including a large survey of registered nurses licensed in New York state, the 2017 American Hospital Association (AHA) Annual Survey, and publicly available 2017 Hospital Compare data from the Centers for Medicare and Medicaid Services (CMS).⁵ Information about patient characteristics and outcomes was derived from CMS MEDPAR data of Medicare patients hospitalized during 2017.

Nurses practicing in hospitals were used as informants about staffing levels and other features of their work environments. The survey of registered nurses was conducted between December 2019 and February 2020. Email addresses of all actively licensed nurses were obtained from the New York state licensure list. All nurses, not a sample, were contacted by email to complete the survey and responses were returned anonymously. Nurses who did not respond to the initial survey invitation received up to 10 follow-up invitations during the study period. Once nurses completed the survey, they no longer received these follow-up invitations; and nurses could opt-out at any time. Nurses were asked to report the name of their hospital employer, which allowed us to aggregate responses from nurses working in the same hospital and create hospital-level measures of nursing resources, such as patient-to-nurse staffing ratios. Additional details of the survey methodology have been reported elsewhere, including results of a non-response second survey revealing no response bias in the variables of interest.¹⁵ The nurse-level response rate was 17% yielding 13,000 responses, an average of 24 registered nurses per hospital working in adult medical surgical units, thus providing reliable estimates of staffing in most acute care general hospitals in New York state.¹⁶

Study Sample of Hospitals and Patients

The analytic sample of hospitals included acute care hospitals in New York state. Hospitals were included in the sample if they had at least 5 registered nurses who responded to the survey and reported working on a medical-surgical unit as a direct care staff nurse. Among the final sample of 116 study hospitals, the average number of nurse respondents was 24 and ranged from 5 to 139 nurses per hospital.

The final patient sample consisted of 52,177 Medicare beneficiaries between the ages of 65 and 99 years old who were discharged from one of the 116 study hospitals between January 1, 2017 and December 31, 2017. To be included in the study sample, patients were required to have a principal diagnosis of sepsis present on admission. ICD-10 codes used to identify sepsis are provided in Appendix 1.

Outcome Variables

The patient outcome variables of interest were in-hospital mortality, 60-day mortality, 60-day readmission, and hospital length of stay. In-hospital mortality was defined as a death occurring during the index admission for sepsis; 60-day mortality was defined as a death occurring either in or outside of the hospital within 60 days of the index admission date. A readmission was identified if a patient was readmitted to a hospital (either the index hospital or some other hospital in our study sample) within 60 days of discharge. Our readmission measure excludes patients who died during the index admission (n=7,773) or who were transferred out to another hospital (n=962). Hospital length of stay was calculated during the index hospitalization as the number of days the patient was hospitalized. Patients with lengths of stay longer than 60 days (n=165) and patients who died during the index admission or who were transferred out to another hospital were excluded.

Predictor Variables

The predictor variables of interest included patient-to-nurse staffing ratios and hospital performance on the sepsis bundle for timely and effective sepsis care (SEP-1). Patient-to-nurse staffing ratios were derived from the survey responses of direct care registered nurses working on medical-surgical units. Nurses were asked to report the number of patients they were assigned during their last shift worked. Responses were averaged among nurses working in the same hospital to create a hospital-level measure of medical-surgical patient-to-nurse staffing.

Hospital performance on timely and effective sepsis care was obtained from CMS Hospital Compare data collected between January 1, 2017 and December 31, 2017. The SEP-1 score is a National Inpatient Quality Measure that began in October 2015 as part of CMS’ quality reporting program.¹⁷ Chart abstraction is used to identify the percentage of patients who received appropriate care for severe sepsis and septic shock. Appropriate care includes interventions such as obtaining lactate measurements, blood cultures, and delivering a broad-spectrum antibiotic within 3 hours of sepsis onset for individuals with severe sepsis. Additionally, patients with septic shock require intravenous fluids within 3 hours of onset, vasopressors within 5 hours, and repeat volume assessments within 6 hours. Hospital SEP-1 scores can range between 0%−100% indicated the percentage of patients who received appropriate care for severe sepsis and septic shock. Although it is not within the clinical scope of bedside nurses to order and initiate the sepsis care bundle, nurses are directly responsible for ensuring timely completion of the relevant diagnostic testing and administration of treatments. Thus, the direct care nurse is a key contributor to a hospital’s performance on the SEP-1 bundle.

Covariates

Hospital covariations were included in the modeling to control for potentially confounding relationships. The AHA survey provided data on hospital size, teaching status, and technology capabilities. Size was defined by the number of inpatient beds and categorized as small (≤ 100 beds), medium (101–250 beds), and large (>250 beds). Teaching status was categorized as non-teaching (no medical trainees), minor teaching (0–4 medical trainees per bed), and major teaching (≥4 medical trainees per bed). Hospitals with the capabilities to perform major organ transplantation and/or open-heart surgery were defined as having a high-technology status. A measure of patient-to-nurse staffing ratios in intensive care units (ICUs) was derived from the survey of nurses and included as a control variable in the models.

Patient covariates were obtained from MEDPAR data and included: age, sex, 32 Elixhauser comorbidities, a dummy indicator for whether or not the patient was a transfer-in from another hospital, and five dummy variables for diagnostic related groups (DRGs), which represented over 98% of the study patients.

Data Analysis

Descriptive statistics including mean, median, standard deviation and ranges are used to report the number of direct care medical-surgical nurse respondents per hospital and the number of Medicare patients with sepsis within the 116 study hospitals. Patient-to-nurse ratios and SEP-1 scores are reported for the 116 study hospitals and the hospitals are described by their distribution of size, teaching status, and technology capabilities. Multi-level random intercept logistic regression models were estimated using MLWin to compute odds ratios for mortality and readmission outcomes. Zero-truncated negative binomial models with clustered robust standard errors were used to compute incident rate ratios (IRRs) for length of stay. Models are first presented unadjusted and then adjusted for hospital and patient covariates. Staffing was modeled as a one-unit change in the number of patients per nurse. SEP-1 score was modeled as a 10-point change in the hospital score.

RESULTS

The numbers of medical-surgical nurse respondents and sepsis patients in the 116 study hospitals are described in Table 1. Among the 116 New York hospitals in our sample, we obtained data about patient-to-nurse staffing ratios from 2,747 registered nurses, with an average of 23.7 nurses per hospital. Data from 52,177 Medicare patients hospitalized with a principal diagnosis of sepsis were included in our analysis, with an average of roughly 450 patients per hospital.

Table 1.

Numbers of Medical-Surgical Nurse Respondents in the 116 Study Hospitals, and the Numbers of Sepsis Patients Used in the Different Analyses

		Nurse Respondents per Hospital
	Number of Nurse Respondents	Mean	SD	Median	Minimum	Maximum
Medical-Surgical Nurses	2,747	23.7	24.3	14	5	139
		Sepsis Patients per Hospital
Sepsis Patients Used in Analyses of -	Number of Patients	Mean	SD	Median	Minimum	Maximum
In-Hospital and 60-Day Mortality	52,177	449.8	403.5	289	11	1,943
Readmissions	43,442	374.5	348.6	232	10	1,699
Length of Stay	43,227	373.1	347.3	229	10	1,696

Characteristics of the study hospitals and the patient-to-nurse staffing ratios and performance on the CMS sepsis bundle (SEP-1) are described in Table 2. The majority of hospitals in our sample had greater than 250 beds (58.6%) and did not have a high technology status (53.5%). Hospitals were relatively evenly distributed by their teaching status: non-teaching (26.7%), minor teaching (37.1%), major teaching (31.0%). The average medical-surgical patient-to-nurse staffing ratio among the 116 hospitals was 6.3 patients per nurse (SD: 1.0). Larger, major teaching hospitals tended to have higher staffing ratios as compared to smaller non-teaching hospitals. The average SEP-1 score was 47.0% (SD 17.5%). SEP-1 scores were higher (indicating better performance on sepsis bundle adherence) in medium-sized hospitals (101–250 beds) and hospitals without high technology capabilities.

Table 2.

Selected Characteristics of the 116 Hospitals in the Study Sample, and Medical-Surgical Staffing and SEP-1 Scores in Hospitals with Different Characteristics

	Number of Hospitals	Percent of Hospitals	Medical-Surgical Staffing (Patients per Nurse)			Severe Sepsis and Septic Shock Management Bundle (SEP-1)
			Mean	SD	Median	Mean	SD	Median
Hospital Size
<=100 Beds	12	10.3%	5.8	0.5	5.7	50.9	17.5	49.0
101–250 Beds	36	31.0%	6.3	1.2	6.1	52.2	18.3	53.0
>250	68	58.6%	6.4	0.9	6.2	43.6	16.6	41.0
Total	116	100.0%	6.3	1.0	6.1	47.0	17.5	46.0
Teaching Status
Non-Teaching	31	26.7%	5.9	0.7	5.8	46.8	17.1	48.0
Minor Teaching	43	37.1%	6.3	1.1	6.1	49.6	16.7	48.0
Major Teaching	36	31.0%	6.5	1.1	6.5	43.4	17.9	42.0
Missing	6	5.2%	6.5	1.2	6.2	52.0	23.7	57.5
Total	116	100.0%	6.3	1.0	6.1	47.0	17.5	46.0
Technology
Non-High Technology	62	53.5%	6.2	0.9	6.1	47.4	17.2	46.5
High Technology	32	27.6%	6.0	0.8	5.9	41.6	15.3	42.5
Missing	22	19.0%	6.9	1.4	6.7	53.9	19.5	58.5
Total	116	100.0%	6.3	1.0	6.1	47.0	17.5	46.0

Among the 52,177 sepsis patients in our sample, 14.9% died during the index admission and 28.6% of patients died within 60-days of admission (Table 3). Excluding individuals who died during the index hospitalization and those who were transferred out, 23.5% were readmitted within 60 days of discharge. The average length of stay during the index hospitalization was 8.5 days (SD 7.3 days). The distribution of patients’ age and sex are described in Table 3. A slightly larger percentage of patients were female (52.3%) as compared to male (47.7%). The most common DRGs were for severe sepsis without mechanical ventilation with (DRG 872) and without (DRG 871) major complication/comorbidity. Common comorbidities included hypertension, fluid and electrolyte disorders, congestive heart failure, and chronic pulmonary disease.

Table 3.

Outcomes and Selected Characteristics of Sepsis Patients in the Study Hospitals

	Patients
Patient Outcomes	Number	Percent
In Hospital Mortality/Cases	7,773/52,177	14.9%
60-Day Mortality/Cases	14,898/52,177	28.6%
60-Day Readmissions/Cases	10,206/43,442	23.5%
	Number	Mean (SD)
Length of Stay	43,277	8.5 (7.3)
Patient Characteristics	Number	Percent
Age
65–69	8,743	16.8%
70–74	8,343	16.0%
75–79	8,628	16.5%
80–84	8,820	16.9%
85–89	8,804	16.9%
90–99	8,839	16.9%
Total	52,177	100.0%
Sex
Female	27,294	52.3%
Male	24,883	47.7%
Total	52,177	100.0%
Transfer Status
Not-Transferred In	50,421	96.6%
Transferred In	1,756	3.4%
Total	52,177	100.0%
Diagnostic-Related Group
853: Infectious Disease with MCC	4,065	7.8%
854: Infectious Disease with CC	805	1.5%
870: Severe Sepsis with MV>96h	2,316	4.4%
871: Severe Sepsis without MV>96h without MCC	34,030	65.2%
872: Severe Sepsis without MV>96h with MCC	10,191	19.5%
Other DRG	770	1.5%
Total	52,177	100%
Common Comorbidities
Hypertension	40,071	76.8%
Fluid and electrolyte disorders	31,960	61.3%
Congestive heart failure	16,640	31.9%
Chronic pulmonary disease	15,718	30.1%
Renal failure	14,746	28.3%
Deficiency Anemias	14,471	27.7%
Diabetes w chronic complications	12,012	23.0%
Other neurological disorders	10,389	19.9%
Hypothyroidism	9,613	18.4%
Diabetes wo chronic complications	7,978	15.3%
Weight loss	7,455	14.3%
Coagulopathy	6,797	13.0%
Depression	6,447	12.4%
Valvular disease	6,273	12.0%
Obesity	6,205	11.9%
Peripheral vascular disease	4,712	9.0%
Paralysis	4,410	8.5%
Metastatic cancer	2,934	5.6%
Solid tumor without metastasis	2,744	5.3%

Notes. MCC: major complication or comorbidity. CC: complication or comorbidity. MV: mechanical ventilation. Readmissions are based on cases that exclude cases that died in the hospital or were transferred to another hospital. Cases used to calculate length of stay exclude cases involving in-hospital deaths, patients transferred to another acute care facility, and lengths of stay longer than 60 days. Comorbidities shown are those that involved at least 5% of the patients in either patient group, ordered according to their prevalence. The percentages of cases with different comorbidities do not sum to 100% due to pa heMs with multi-comorbidities.

Table 4 presents the unadjusted and adjusted effects of nurse staffing and SEP-1 scores on clinical outcomes of sepsis patients. In the unadjusted models, we find large and significant effects of staffing on all four clinical outcomes of interest. Each additional patient per nurse is associated with 19% higher odds of in-hospital mortality (OR 1.19, 95% CI 1.10–1.29, p<0.0001), 13% higher odds of 60-day mortality (OR 1.13, 95% CI 1.07–1.19, p<0.0001), 6% higher odds of readmission (OR 1.06, 95% CI 1.02–1.10, p 0.004), and longer lengths of stay by a factor of 6% (IRR 1.06, 95% CI 1.02–1.11, p=0.009). In the unadjusted models, a 10-point increase in a hospital’s performance on the SEP-1 bundle was significantly associated with shorter lengths of stay by a factor of 3% (IRR 0.97, 95% CI 0.95–1.00, p=0.017), but was not significantly associated with lower mortality or readmission.

Table 4.

Unadjusted and Adjusted Effects of Medical Surgical Staffing (Patient-to-Nurse Ratio) and SEP-1 Scores on Sepsis Patient Outcomes

		Unadjusted Models		Fully Adjusted Models
Patient Outcome		Staffing Effect	SEP-1 Effect	Staffing Effect	SEP-1 Effect
In-Hospital Mortality	Odds Ratio	1 19***	0.96	1.12**	0.95*
	95% CI	(1.10, 1.29)	(0.91,1.00)	(1.03, 1.21)	(0.91, 0.99)
	P>|z|	<0.0001	0.051	0.008	0.018
60-Day Mortality	Odds Ratio	1.13***	0.99	1.07*	0.97
	95% CI	(1.07, 1.19)	(0.96, 1.02)	(1.01, 1.14)	(0.94, 1.00)
	P>|z|	<0.0001	0.40	0.028	0.056
60-Day Readmission	Odds Ratio	1.06**	0.99	1.07**	0.99
	95% CI	(1.02, 1.10)	(0.96, 1.01)	(1.03, 1.12)	(0.97, 1.02)
	P>|z|	0.004	0.179	0.001	0.613
Length of Stay	IRR	1.06**	0.97*	1.05**	0.98*
	95% CI	(1.02, 1.11)	(0.95, 1.00)	(1.02, 1.09)	(0.97, 1.00)
	P>|z|	0.009	0.017	0.002	0.024

Notes.

^*p<0.05

^**p<0.01

^***p<0.001.

Odds ratios for mortality and readmission models are from random intercept models estimated using MLWin. Incident rate ratios (IRR) for length of stay models are from zero truncated negative binomial models with clustered standard errors. In the adjusted models for all outcomes, hospital-level controls include hospital size, technology status, teaching status, and ICU staffing, and patient controls include age, sex, transfer status, 32 Elixhauser comorbidities, and dummy variables for the different DRGs.

The fully adjusted models jointly modeled both the staffing and SEP-1 score effects, in addition to hospital-level and patient-level covariates. The effects of each additional patient per nurse in a nurse’s workload remained large and significant for all of the clinical outcomes studied. A 10-point improvement in SEP-1 scores was significantly associated with shorter lengths of stay and 5% lower odds of in-hospital mortality (OR 0.95, 95% CI 0.91–0.99, p=0.018).

DISCUSSION

This analysis demonstrates that patient-to-nurse staffing ratios on medical-surgical units vary across hospitals in New York – a state currently considering hospital nurse staffing regulation – with the average medical-surgical nurse caring for 2.3 more patients than is recommended under the proposed legislation of 4 patients per nurse.¹⁴ In this study, we found each additional patient in a nurse’s workload to be strongly and significantly associated with a higher probability of in-hospital and 60-day mortality, readmission, as well as longer lengths of stay, even after accounting for hospital and patient characteristics.

There have been substantial policy efforts over the last decade to reduce mortality among sepsis patients – both through Rory’s Regulations in the case of New York state and through CMS sepsis bundles that apply to hospitals nationally. This study finds that patients in hospitals with greater adherence to the SEP-1 sepsis bundle have lower in-hospital mortality and somewhat shorter lengths of stay. No significant relationships were found between hospital adherence to the SEP-1 bundle and 60-day mortality or readmissions.

Notably, the effects of nurse staffing on patient outcomes are more pronounced than is hospital adherence to the SEP-1 bundle. For example, each additional patient per nurse is associated with 12% higher odds of in-hospital mortality compared with a 10% change in SEP-1 adherence associated with only a 5% change in in-hospital mortality. Higher SEP-1 scores were also associated with shorter lengths of stay, but staffing had more than twice as large an effect on shorter lengths of stay, even when accounting for hospitals’ SEP-1 scores. Moreover, the effect of staffing was large and significant in terms of 60-day mortality and readmissions, while the SEP-1 scores revealed no association.

Nurse staffing levels have not been previously studied in relation to evaluations of the impact of the SEP-1 bundle. However, it is not entirely surprising to find nurse staffing workloads are associated with sepsis outcomes, above and beyond hospitals’ adherence to the SEP-1 bundles. The interventions comprising the SEP-1 bundle and the overall care of a septic patient are heavily reliant on nurses with the adequate time and resources to surveil for signs and progression of sepsis, to obtain blood samples in a timely manner, and to administer antibiotic and vasopressor medication and fluid resuscitation, which requires close monitoring and titration.

The findings suggest that sepsis patient outcomes would likely be substantially improved by establishing a minimum safe hospital nurse staffing standard, like the one currently under consideration in New York state, in addition to the policies such as Rory’s Regulations to promote adherence to the SEP-1 bundles. Attention to nurse staffing ratios may not only reduce mortality and readmission among sepsis patients as we show here but is likely to impact patients with a wide range of medical and surgical conditions, as previous research has suggested.^18–22

Limitations

The study findings should be considered in the context of both its strengths and limitations. While this was a cross-sectional design, and can therefore not claim a causal relationship between nurse staffing and patient outcomes, other studies using multiple waves of panel data have shown nurse staffing to be associated with patient outcomes over time.²³ Our study uses a measure of nurse staffing derived from staff nurses providing direct clinical care on medical-surgical units in a large and representative sample of New York hospitals. Other studies of nurse staffing often rely on measures of staffing created by administrators and include nurse positions in direct as well as indirect patient care roles and in ambulatory as well as inpatient care, which creates a less precise measure of the workload for nurses at the bedside. Finally, we rely on publicly available data from Hospital Compare of hospital-level adherence to the SEP-1 bundle to understand whether and to what extent sepsis patients receive appropriate and timely care. The SEP-1 bundle uses an “all or nothing” approach, such that for a hospital to receive credit for administering appropriate and timely sepsis care, they need to have performed all the interventions within the bundle. Hospitals that provide some of the interventions are not given credit for those interventions.

Conclusions

Despite public attention to high and potentially avoidable deaths resulting from a diagnosis of sepsis as well as resulting policies promoting adherence to evidence-based sepsis care measures, the average hospital provides appropriate sepsis care to only a little more than half of patients. Nurses are central to the recognition, management, and treatment of sepsis; and thus, the resultant clinical outcomes for older adults with sepsis are associated with hospital nurses’ workloads. In this study, we find that every additional patient in a nurses’ workload is associated with higher odds of death, as well as higher odds of readmission and longer lengths of stay, which suggests attention to patient-to-nurse staffing ratios may be critical to increasing adherence to sepsis care and improving patient outcomes.

HIGHLIGHTS.

• New York state was one of the first to require implementation of sepsis bundles.

• Patient-to-nurse staffing ratios vary considerably across hospitals in New York.

• Sepsis outcomes vary despite bundle requirements because of nurse staffing.

• Requiring minimum nurse staffing along with sepsis bundles may improve outcomes.

Appendix 1.

ICD-10 Principal Diagnosis Codes for Sepsis

ICD-10	Description
A021	Salmonella sepsis
A227	Anthrax sepsis
A267	Erysipelothrix sepsis
A327	Listerial sepsis
A400	Sepsis due to streptococcus, group A
A401	Sepsis due to streptococcus, group B
A403	Sepsis due to Streptococcus pneumoniae
A408	Other streptococcal sepsis
A409	Streptococcal sepsis, unspecified
A4101	Sepsis due to Methicillin susceptible Staphylococcus aureus
A4102	Sepsis due to Methicillin resistant Staphylococcus aureus
A411	Sepsis due to other specified staphylococcus
A412	Sepsis due to unspecified staphylococcus
A413	Sepsis due to Hemophilus influenzae
A414	Sepsis due to anaerobes
A4150	Gram-negative sepsis, unspecified
A4151	Sepsis due to Escherichia coli [E. coli]
A4152	Sepsis due to Pseudomonas
A4153	Sepsis due to Serratia
A4159	Other Gram-negative sepsis
A4181	Sepsis due to Enterococcus
A4189	Other specified sepsis
A419	Sepsis, unspecified organism
A427	Actinomycotic sepsis
A5486	Gonococcal sepsis
B377	Candidal sepsis
R6520	Severe sepsis without septic shock
R6521	Severe sepsis with septic shock
R7881	Bacteremia