Probability of sepsis after infection consultations in primary care in the United Kingdom in 2002–2017: Population-based cohort study and decision analytic model

Martin C Gulliford; Judith Charlton; Joanne R Winter; Xiaohui Sun; Emma Rezel-Potts; Catey Bunce; Robin Fox; Paul Little; Alastair D Hay; Michael V Moore; Mark Ashworth; SafeAB Study Group

doi:10.1371/journal.pmed.1003202

. 2020 Jul 23;17(7):e1003202. doi: 10.1371/journal.pmed.1003202

Probability of sepsis after infection consultations in primary care in the United Kingdom in 2002–2017: Population-based cohort study and decision analytic model

Martin C Gulliford ^1,^2,^*, Judith Charlton ¹, Joanne R Winter ¹, Xiaohui Sun ¹, Emma Rezel-Potts ^1,², Catey Bunce ^1,², Robin Fox ³, Paul Little ⁴, Alastair D Hay ⁵, Michael V Moore ⁴, Mark Ashworth ¹; SafeAB Study Group^¶

Editor: Andrew Carson-Stevens⁶

PMCID: PMC7377386 PMID: 32702001

Abstract

Background

Efforts to reduce unnecessary antibiotic prescribing have coincided with increasing awareness of sepsis. We aimed to estimate the probability of sepsis following infection consultations in primary care when antibiotics were or were not prescribed.

Methods and findings

We conducted a cohort study including all registered patients at 706 general practices in the United Kingdom Clinical Practice Research Datalink, with 66.2 million person-years of follow-up from 2002 to 2017. There were 35,244 first episodes of sepsis (17,886, 51%, female; median age 71 years, interquartile range 57–82 years). Consultations for respiratory tract infection (RTI), skin or urinary tract infection (UTI), and antibiotic prescriptions were exposures. A Bayesian decision tree was used to estimate the probability (95% uncertainty intervals [UIs]) of sepsis following an infection consultation. Age, gender, and frailty were evaluated as association modifiers. The probability of sepsis was lower if an antibiotic was prescribed, but the number of antibiotic prescriptions required to prevent one episode of sepsis (number needed to treat [NNT]) decreased with age. At 0–4 years old, the NNT was 29,773 (95% UI 18,458–71,091) in boys and 27,014 (16,739–65,709) in girls; over 85 years old, NNT was 262 (236–293) in men and 385 (352–421) in women. Frailty was associated with greater risk of sepsis and lower NNT. For severely frail patients aged 55–64 years, the NNT was 247 (156–459) in men and 343 (234–556) in women. At all ages, the probability of sepsis was greatest for UTI, followed by skin infection, followed by RTI. At 65–74 years, the NNT following RTI was 1,257 (1,112–1,434) in men and 2,278 (1,966–2,686) in women; the NNT following skin infection was 503 (398–646) in men and 784 (602–1,051) in women; following UTI, the NNT was 121 (102–145) in men and 284 (241–342) in women. NNT values were generally smaller for the period from 2014 to 2017, when sepsis was diagnosed more frequently. Lack of random allocation to antibiotic therapy might have biased estimates; patients may sometimes experience sepsis or receive antibiotic prescriptions without these being recorded in primary care; recording of sepsis has increased over the study period.

Conclusions

These stratified estimates of risk help to identify groups in which antibiotic prescribing may be more safely reduced. Risks of sepsis and benefits of antibiotics are more substantial among older adults, persons with more advanced frailty, or following UTIs.

In this cohort study, Martin Gulliford and colleagues investigate probability of sepsis in primary care patients with or without prescribed antibiotics.

Author summary

Why was this study done?

Sepsis is a severe reaction to an infection that may lead to life threatening damage to organ systems. Sepsis is an increasingly recognised concern for health professionals and patients in primary care.
Inappropriate and unnecessary antibiotic prescribing is a widespread problem in primary care that may be contributing to antimicrobial resistance.
This study aimed to estimate the probability of a patient developing sepsis after an infection consultation in primary care if antibiotics are or are not prescribed.

What did the researchers do and find?

We analysed the electronic health records of all registered patients at 706 general practices, with 66.2 million person-years of follow-up from 2002 to 2017 and 35,244 first episodes of sepsis.
We found that the probability of sepsis was lower if an antibiotic was prescribed, but the number of antibiotic prescriptions required to prevent one episode of sepsis (number needed to treat [NNT]) decreased with age.
Frailty was associated with greater risk of sepsis and lower NNT.
At all ages, the probability of sepsis was greatest for urinary tract infection, followed by skin infection, followed by respiratory tract infection.

What do these findings mean?

These results show that risks of sepsis and benefits of antibiotics are more substantial among older adults, persons with more advanced frailty, or following urinary tract infections.
Antibiotic use may be more safely reduced in groups with lower probability of sepsis.
We caution that our results represent averages over diverse localities and years of study, and lack of random allocation to antibiotic therapy might have caused bias.

Introduction

The threat of antimicrobial drug resistance (AMR) is attracting the concern of national governments and international organisations [1]. Antibiotic-resistant infections are increasing and are more often identified in primary care as well as hospital settings. In the UK, antibiotic prescribing in primary care accounts for more than three-quarters of all antibiotic use. Respiratory tract infections (RTIs) represent the most common reason for antibiotic treatment [2], with general practitioners prescribing antibiotics at about half of the consultations for ‘self-limiting’ RTIs, including common colds, acute cough and bronchitis, sore throat, otitis media, and rhinosinusitis [3], with little change over the last 2 decades [4,5]. The other main indications for antibiotic prescription include urinary tract infections (UTIs) and skin infections [2,6,7]. The UK government has developed a 5-year antimicrobial resistance strategy that identifies reducing unnecessary antibiotic prescribing and improving antibiotic selection as key elements of antimicrobial stewardship [8,9].

Reducing antibiotic use might potentially compromise patient safety by increasing the risk of serious bacterial infections following consultations for common infections [10]. The safety of reduced antibiotic prescribing is a major concern for both clinicians and patients [11]; parents may also be particularly concerned about safety issues, which are often an important motivation for seeking active treatment for children [12]. A systematic review of qualitative studies found that clinicians commonly prescribe antibiotics ‘just in case’ they might be needed [13]. Based on international comparisons, with both low [14] and high [15] antibiotic prescribing being observed across Europe without apparent risks to patient safety, it appears that a substantial reduction of antibiotic prescribing in primary care might be reasonable. However, only a few existing research studies directly address the safety outcomes of reduced antibiotic prescribing at consultations for common infections in primary care.

Strategies to reduce inappropriate use of antibiotics must ensure that antibiotics can be used when they are needed [16,17]. Bacterial infections are still of public health importance, and there has been growing recognition of the importance of sepsis, with more than 200,000 hospital admissions for sepsis each year in England and up to 59,000 deaths [18]. Early recognition and treatment of sepsis is being promoted by health services and professional organisations through assessment of risk for individual patients [19]. In the UK, a National Early Warning Score (NEWS2) based on six physiological parameters has been promoted to identify individual patients who may be at risk of sepsis [20]. However, this approach has also been criticised because early warning signs of sepsis are often nonspecific, and alerting systems may result in false-positive signals at many consultations [21].

Research is needed to provide quantitative estimates of risk that might provide clinicians and patients with evidence to inform antibiotic prescribing decisions. This study aimed to estimate the probability of sepsis if antibiotics were prescribed or not and to estimate the number of antibiotic prescriptions required to prevent one episode of sepsis. We estimated the probability of sepsis for groups of patients characterised by age, gender, and frailty as well as reason for consultation.

Methods

Ethics statement

Scientific and ethical approval of the protocol was given by the Clinical Practice Research Datalink (CPRD) Independent Scientific Advisory Committee (ISAC protocol 18-041R). The study was based on analysis of fully anonymised data, and individual consent was not required.

Data source

We carried out a population-based cohort study in the UK CPRD GOLD database, employing data for 2002–2017. The CPRD GOLD (www.cprd.com) is one of the world’s largest databases of primary care electronic health records, with participation of about 7% of UK family practices and with ongoing collection of anonymised data from 1990 [22]. CPRD GOLD is considered to be geographically and sociodemographically representative of the UK population [22]. The high quality of CPRD GOLD data has been confirmed in many studies [23]. The protocol for the study has been published (https://fundingawards.nihr.ac.uk/award/16/116/46). Descriptive data for antibiotic prescribing and general practice–level associations have been reported previously [24]. This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 STROBE Checklist).

Sepsis events

We ascertained sepsis events from the entire registered population of CPRD GOLD because these are generally rare events. Incident cases of sepsis were obtained from CPRD GOLD for the years 2002–2017, with person-years at risk providing the denominator. The start of the patient record was the latest of 1 year after the patient’s current registration date, the date the general practice began contributing up-to-standard data to CPRD GOLD, or 1 January 2002. The end of the patient’s record was defined as the earliest of the end of registration, the patient’s death date, or 31 December 2017. The mean duration of follow-up was 6.9 years. Sepsis events were evaluated using Read codes recorded into patients’ clinical and referral records [24]. There were 77 Read codes for sepsis and septicaemia, but the four most frequent codes accounted for 92% of events including ‘Sepsis’ (two codes), ‘Septicaemia’, and ‘Urosepsis’ (S1 Table). We included incident first events in further analyses; recurrent events in the same patient were not evaluated further because it may not always be possible to distinguish new occurrences from reference to ongoing or previous problems in electronic health records.

For each sepsis event, we evaluated whether a consultation for a common infection was recorded within the preceding 30 days. We employed a 30-day time window with the intention of capturing data for acute infections and their short-term outcomes. We identified consultations for RTIs (including upper and lower RTIs), skin infections, and UTIs (including cystitis and uncomplicated UTIs only) because these are the most important groups of conditions for which antibiotics are prescribed in primary care [25] (S2 Table). We evaluated Read codes in patients’ clinical and referral records in order to identify consultations associated with common infections. We also evaluated whether an antibiotic prescription was issued during the 30 days preceding a sepsis event, either on the same date as an infection consultation or on a different date [24,25] (S3 Table).

Selection of sample for antibiotic prescribing analysis

We estimated infection consultation rates and the proportion of consultations with antibiotics prescribed from a sample of patients registered with CPRD GOLD. This was because it is not feasible to download and analyse data for the millions of records represented by all infection consultations and antibiotic prescriptions over 16 years [24]. A random sample of patients was drawn from the list of all registered patients, stratifying by year between 2002 and 2017 and by family practice. The ‘sample’ command in the R programme was employed to provide a computer-generated random sequence. In each year of study, a sample of 10 patients was taken for each gender and age group using 5-year age groups up to a maximum of 104 years. Each sampled patient contributed data in multiple years of follow-up. There was a total sample of 671,830 individual patients registered at a total of 706 family practices who contributed person-years between 2002 and 2017. The sampling design enabled estimation of all age-specific rates with similar precision, and age-standardisation provided weightings across age groups. Data for antibiotic prescribing in this sample have been reported previously [24] (S4 Table).

For each patient in the antibiotic prescribing sample, we calculated the person-years at risk between the start and end of the patient’s record. Person-years was grouped by gender, age group, and comorbidity. Age groups were from 0 to 4, 5 to 9, and 10 to 14 years and then 10-year age groups up to 85 years and over. Infection consultations were evaluated using Read codes as outlined above. Antibiotic prescriptions were evaluated using product codes for antibiotics listed in section 5.1 of the British National Formulary, excluding methenamine and drugs for tuberculosis and leprosy. Different antibiotic classes and antibiotic doses were not considered further in this analysis. Multiple antibiotic prescription records on the same day were considered as a single antibiotic prescription.

Evaluation of frailty

We used Clegg’s e-Frailty Index to evaluate frailty level [26]. The e-Frailty Index includes 36 deficits, which are evaluated as present or absent based on Read-coded electronic health records. Patients were classified as being ‘nonfrail’ or having ‘mild’, ‘moderate’, or ‘severe’ frailty based on the number of deficits recorded. We evaluated frailty for each patient in each calendar year of study [27] in order to provide a frailty estimate for the index year of each sepsis episode. We also estimated consultation rates and antibiotic prescribing proportions by frailty category for the antibiotic prescribing sample. As full electronic health record data were not available for the entire CPRD GOLD denominator, we allocated person-years to frailty categories using the proportion in each frailty category that we observed in the antibiotic prescribing sample. Although the concept of frailty may be applied at any age, frailty was only evaluated from 55 years and older because most patients under the age of 55 years were classed as nonfrail or as having only mild frailty (S5 Table).

Decision tree

In order to evaluate the probability of sepsis following an infection consultation in primary care, we constructed a decision tree (Fig 1) [28]. An individual developing an infection may decide to consult their general practice or not; if they consult, they may be prescribed antibiotics or not; subsequently, they may develop sepsis or not. We used estimates from CPRD data analysis to populate the decision tree with empirical estimates for probabilities as outlined in Table 1. We used Bayes’ theorem to estimate the probability of sepsis following an infection consultation if antibiotics were prescribed or if antibiotics were not prescribed. We estimated the ‘number needed to treat’ (NNT), the number of antibiotic prescriptions required to prevent one sepsis event, as the reciprocal of the difference in probability of sepsis with and without antibiotics. We obtained central estimates and 95% uncertainty intervals from 10,000 random draws from the beta distribution [29]. All estimates were stratified by gender and 10-year age group. For the population aged 55 years and older, we also stratified by frailty category. We also evaluated subgroups of common infections, including RTI, skin infections, and UTI.

Table 1. Definition and data source for probabilities.

Term	Explanation	Data source
P(Infection)	Probability of a person consulting with infection in a 30-day period	From infection consultation rate per 30 days in sampled data set from CPRD
P(AB \| Infection)	Probability of receiving an AB prescription on the same date as an infection consultation	From proportion of infection consultations with AB prescribed in sampled data set from CPRD
P(Sepsis)	Probability of sepsis, per 30 days	From incidence of sepsis from entire registered CPRD population
P(Infection \| Sepsis)	Probability of patients with sepsis having consulted for an infection in 30 days preceding their sepsis diagnosis	Proportion of sepsis cases with previous infection consultation, calculated from entire registered CPRD population
P(Sepsis \| Infection)	Probability of sepsis in the 30 days following an infection consultation	$\frac{P (I n f e c t i o n \| S e p s i s) P (S e p s i s)}{P (I n f e c t i o n)}$
P(Sepsis \| [AB \| Infection])	Probability of sepsis having consulted for an infection and received an AB prescription	$\frac{P ([A B \| Infection] \| Sepsis) P ([Sepsis \| Infection])}{P (A B \| Infection)}$
P(Sepsis \| [NoAB \| Infection])	Probability of sepsis having consulted for an infection and not received an AB prescription	$\frac{P ([NoAB \| Infection] \| Sepsis) P ([Sepsis \| Infection])}{P (NoAB \| Infection)}$
NNT	The number of additional antibiotic prescriptions required to prevent one case of sepsis	$\frac{1}{P (S e p s i s \| [A B \| I n f e c t i o n]) - P (S e p s i s \| [N o A B \| I n f e c t i o n])}$

Open in a new tab

Abbreviations: AB, antibiotic; CPRD, Clinical Practice Research Datalink; NNT, number needed to treat; P, probability

Sensitivity analyses

In sensitivity analyses, we evaluated whether use of a 60-day time window gave different results from a 30-day time window. The primary analysis reported data for a 16-year period, but the incidence of recorded sepsis has been increasing [24]. We repeated the analysis using only data for 4-year periods from 2002–2005 to 2014–2017 to evaluate whether estimates differed from the whole period from 2002 to 2017. We also investigated whether estimates differed if sepsis diagnoses recorded in Hospital Episode Statistics (HES) or as causes of death on mortality certificates were included. The sample for linkage was obtained from CPRD (Linkage Set 16). The linked sample included data for 378 English general practices, with 5,524,983 patients providing primary care electronic records data linked to HES and mortality statistics. We searched for ICD-10 codes for sepsis and septicaemia. We included primary diagnoses from HES-admitted patient care records and all mentions of sepsis in mortality statistics data. We repeated analyses using primary care electronic health records alone, primary care electronic health records with linked HES data, or primary care electronic health records with linked HES and mortality data.

Results

The study included 706 general practices, with a total of 66.2 million person-years of follow-up (S1 Fig). Data for the distribution of sepsis patients by age and gender are shown in Table 2; data by region and period are shown in S3 Table. The probability of a consultation with a common infection of the skin, RTI, or UTI in any 30-day period ranged between 0.02 (1 in 50) and 0.08 (1 in 12). This probability of an infection consultation was higher in children and old people and greater in women than men during midlife (Tables 2 and 3). The probability of an antibiotic being prescribed at an infection consultation ranged between 0.43 and 0.67, with the probability being lowest for young children in whom consultation rates are highest (Table 3).

Table 2. First sepsis events recorded in CPRD from 2002 to 2017 and preceding infection consultations and AB prescriptions.

Gender	Age group (years)	Sepsis events	Infection consultations in previous 30 days	Proportion (%) of sepsis events preceded by infection consultations	AB prescriptions on same date	Proportion (%) of infection consultations with ABs prescribed
Male	0–4	224	51	22.8	11	21.6
	5–14	303	48	15.8	6	12.5
	15–24	360	59	16.4	21	35.6
	25–34	449	78	17.4	18	23.1
	35–44	791	117	14.8	24	20.5
	45–54	1,342	241	18.0	47	19.5
	55–64	2,466	472	19.1	102	21.6
	65–74	3,933	724	18.4	155	21.4
	75–84	4,752	1,089	22.9	256	23.5
	85+	2,738	713	26.0	158	22.2
Female	0–4	204	55	27.0	12	21.8
	5–14	238	32	13.4	9	28.1
	15–24	500	76	15.2	24	31.6
	25–34	806	110	13.6	38	34.5
	35–44	1,095	175	16.0	41	23.4
	45–54	1,631	267	16.4	72	27.0
	55–64	2,443	445	18.2	119	26.7
	65–74	3,215	646	20.1	180	27.9
	75–84	3,982	890	22.4	204	22.9
	85+	3,772	984	26.1	222	22.6

Open in a new tab

Abbreviations: AB, antibiotic; CPRD, Clinical Practice Research Datalink

Table 3. Probability of sepsis after infection consultations in primary care.

		Probability of. . .
		Infection consultation per 30 days	First sepsis event in any 30-day period	Infection consultation 30 days before sepsis event	AB at infection consultation	Sepsis after infection consultation, no AB	Sepsis after infection consultation, AB
Gender	Age (years)	P(Infection)	P(Sepsis)	P(Infection \| Sepsis)	P(AB \| Infection)	P(Sepsis \| [No AB \| Infection])	P(Sepsis \| [AB \| Infection])	NNT (95% UI)
Male	0–4	0.08	0.000014	0.23	0.43	0.000054	0.000020	29,773 (18,458–71,091)
	5–14	0.04	0.000006	0.16	0.48	0.000047	0.000008	25,606 (17,962–40,817)
	15–24	0.02	0.000008	0.17	0.58	0.000101	0.000041	16,921 (10,285–39,551)
	25–34	0.02	0.000009	0.17	0.60	0.000193	0.000039	6,517 (4,779–9,522)
	35–44	0.02	0.000013	0.15	0.62	0.000239	0.000039	5,035 (3,980–6,610)
	45–54	0.02	0.000022	0.18	0.62	0.000472	0.000071	2,497 (2,121–2,999)
	55–64	0.02	0.000048	0.19	0.63	0.000825	0.000135	1,449 (1,282–1,652)
	65–74	0.03	0.000105	0.18	0.64	0.001305	0.000202	907 (823–1,007)
	75–84	0.04	0.000219	0.23	0.63	0.002700	0.000478	450 (413–492)
	85+	0.05	0.000416	0.26	0.61	0.004647	0.000833	262 (236–293)
Female	0–4	0.08	0.000014	0.27	0.43	0.000060	0.000023	27,014 (16,739–65,709)
	5–14	0.04	0.000005	0.14	0.51	0.000025	0.000010	65,522 (35,239–240,067)
	15–24	0.04	0.000012	0.15	0.61	0.000080	0.000024	18,120 (12,472–30,241)
	25–34	0.04	0.000016	0.14	0.63	0.000105	0.000033	13,926 (10,044–21,273)
	35–44	0.04	0.000018	0.16	0.66	0.000184	0.000030	6,513 (5,349–8,194)
	45–54	0.03	0.000028	0.16	0.66	0.000278	0.000054	4,463 (3,756–5,421)
	55–64	0.04	0.000048	0.18	0.67	0.000490	0.000088	2,486 (2,179–2,876)
	65–74	0.04	0.000080	0.20	0.67	0.000793	0.000151	1,557 (1,388–1,758)
	75–84	0.05	0.000138	0.22	0.66	0.001525	0.000231	773 (705–847)
	85+	0.05	0.000271	0.26	0.64	0.003110	0.000509	385 (352–421)

Open in a new tab

Abbreviations: AB, antibiotic; NNT, number needed to treat; P, probability; UI, uncertainty interval

There were 35,244 first episodes of sepsis between 2002 and 2017. The probability of an infection consultation within 30 days before a sepsis event ranged between 0.14 (1 in 7) and 0.26 (1 in 4), with higher values at the extremes of age (Table 3). If no antibiotic was prescribed, the probability of sepsis at age 0–4 years was 0.000054 (1 in 18,519 consultations) in males and 0.000060 (1 in 16,667) in females. The probability of sepsis following an infection consultation without antibiotics increased linearly with age on a log scale (Fig 2, upper panel), reaching 0.004647 (1 in 215 consultations) in males and 0.003110 (1 in 321 consultations) in females aged 85 years and older (Table 3). If antibiotics were prescribed at an infection consultation, the estimated probability of sepsis was lower, ranging from 0.000020 (1 in 50,000 consultations) in males and 0.000023 (1 in 43,478 consultations) in females at age 0–4 years and to 0.000833 (1 in 1,200 consultations) in males and 0.000509 (1 in 1,965 consultations) in females aged 85 years and older. The number of antibiotic prescriptions required to prevent one sepsis event was highly age dependent (Fig 2, lower panel). For children aged 0–4 years, the NNT was 29,773 (18,458–71,091) in males and 27,014 (16,739–65,709) in females. However, at age 85 years and older, the NNT was 262 (236–293) in males and 385 (352–421) in females.

Fig 2 — Number of antibiotic prescriptions required to prevent one sepsis event (NNT) (lower panel). Figures are median probabilities (95% uncertainty interval). AB, antibiotic; NNT, number needed to treat.

In the population aged 55 years and older, estimates were obtained separately by frailty category (Fig 3, S7 Table). The probability of sepsis was greater, and the NNT smaller, for patients with more advanced frailty. For nonfrail patients aged 65–74 years, the NNT was 1,680 (1,354–2,133) for men and 2,718 (2,089–3,697) for women. But for patients of the same age with severe frailty, the NNT was 259 (196 to 360) for men and 438 (329 to 624) for women. For patients with severe frailty, the NNT was less than 250 in men and less than 400 in women for all age groups over 55 years. For nonfrail patients, the probability of sepsis increased, and the NNT decreased, with increasing age (Fig 3). In nonfrail patients, the NNT declined from 2,309 (1,890–2,879) in men and 3,782 (3,001–4,907) in women at 55–64 years old to 407 (274–677) in men and 499 (346–780) for women at 85 years and older. Estimates for patients with mild or moderate frailty exhibited an intermediate pattern (Fig 3).

The probability of sepsis was higher following consultations for UTI than for skin infections or RTI, a pattern of association that was observed across all age groups and men and women (Fig 4, S8 Table). For patients aged 65 without antibiotic treatment, the probability of sepsis following an RTI consultation was 0.00090 (1 in 1,111 consultations) in men and 0.00053 (1 in 1,887 consultations) in women; following a skin infection consultation, the probability was 0.00224 (1 in 446) in men and 0.00150 (1 in 667) in women; following a UTI consultation, the probability was 0.009227 (1 in 108) in men and 0.003787 (1 in 264) in women. At the same age, the corresponding numbers needed to treat were as follows: for RTI, the NNT for men was 1,257 (1,112–1,434), and the NNT for women was 2,278 (1,965–2,686); for skin infection, the NNT for men was 502 (398–646), and the NNT for women was 784 (602–1,051); for UTI consultations, the NNT for men was 120 (102–145), and the NNT for women was 284 (241–342) (Fig 4).

Fig 4 — Figures are median estimates (95% uncertainty interval). Uncertainty intervals were omitted for 0–4 years and 5–9 years if data were too sparse to give reliable estimates. NNT, number needed to treat; RTI, respiratory tract infection; UTI, urinary tract infection.

Sensitivity analyses

An analysis employing a 60-day time window to evaluate exposure gave generally similar results to those using a 30-day time window. In men aged 85 and over, the NNT for all infections was 262 (236–293) with a 30-day time window but 313 (276–359) with a 60-day window; for women of the same age, the figures were 385 (352–421) and 466 (419–523), respectively. When the analysis results were compared for the 4-year periods from 2002–2005 to 2014–2017, estimates for the probability of sepsis were slightly higher, and NNT slightly lower, for the most recent period (S2 Fig), consistent with the higher reported incidence of sepsis in this period (S9 Table). In the oldest age group, from 85 years and over, the probability of sepsis without antibiotics was as follows: for 2014–2017, the probability for men was 0.007287, and the probability for women was 0.004775; with antibiotics, the probability for men was 0.001290, and the probability for women was 0.000839; the NNT for men was 167 (141–202), and the NNT for women was 254 (216–302).

In the linked sample, there were 42,785 first sepsis events across all three data sources, including 17,341 from primary care records, 17,363 from HES admitted patient care (APC) primary diagnoses, and 8,081 from Office for National Statistics (ONS) mortality records during 36.2 million patient-years of follow-up. Accordingly, the underlying probability of sepsis was greater when linked records were employed. However, sepsis events in HES and ONS mortality statistics were less frequently associated with preceding infection consultations in general practice (S3 Fig). Consequently, the probability of sepsis following an infection consultation was only slightly higher if linked data were included in the analysis (S4 Fig), and the estimated NNT was only slightly lower (S5 Fig).

Discussion

Main findings

This study analysed primary care electronic health records data for a large population followed for 16 years with 35,244 new sepsis events. We found that the probability of sepsis following consultation for common infection episodes in primary care is highly age dependent. Without antibiotic treatment, sepsis may follow less than 1 in 10,000 infection consultations under 25 years of age and less than 1 in 1,000 consultations under 65 years of age. The probability of sepsis increases at older ages, and sepsis may follow approximately 1 in 200 (men) or 1 in 300 (women) consultations at age 85 or older. At older ages, the probability of sepsis is also highly dependent on frailty level: 55-year-olds with severe frailty have a similar probability of sepsis as a nonfrail 85-year-old. The probability of sepsis is related to infection type, with the greatest probability following consultations for UTI, the least for RTI, and consultations for skin infections being in an intermediate position. Risks were generally slightly higher for men, which might be accounted for by their generally lower consultation rates.

The incidence of recorded sepsis has been increasing over time with more inclusive case definitions and increasing awareness of the condition [24,30]. When we estimated the main results for the period from 2014 to 2017, the probability of sepsis was higher and NNT lower than for the period from 2002 to 2017. Although we caution that the absolute values of estimates may vary depending on the temporal or geographical context, we expect that in relative terms estimates will continue to identify older age, frailty, and UTI consultations as being associated with greatest risks of sepsis.

Sepsis is an uncommon but concerning outcome of common infection episodes in primary care. Appropriate antibiotic therapy may have immediate benefits that are not restricted to reduction in risk of sepsis, but antibiotic prescriptions are also often associated with immediate harms in the form of drug side effects. The potential risk of antimicrobial resistance has a significance that extends beyond the context of an individual consultation. Prescribing decisions must therefore be informed by the balance of all of the benefits and harms of either prescribing or not prescribing antibiotics. Quantification of the possible risks of sepsis contributes to informing these decisions.

Strengths and limitations

The study drew on a large population-based cohort, enabling us to analyse representative data and obtain precise estimates that may be widely applicable. However, electronic health records comprise clinical data with several limitations and potential for bias. We analysed the outcomes of clinical decisions on whether to prescribe antibiotics or not. In the absence of randomisation, it may be expected that antibiotics were prescribed to individuals at higher risk, whereas lower-risk patients may be less likely to be prescribed antibiotics. Consequently, the probability of sepsis may be underestimated (in comparison with a study employing random allocation) in the absence of antibiotics and overestimated for patients receiving antibiotics, with the NNT being overestimated. However, the analysis depended on general practice electronic health records of antibiotic prescriptions, which account for about 85% of community antibiotic prescribing [2], but we cannot exclude the possibility that patients might have obtained antibiotic prescriptions from alternative sources, including out-of-hours services. Measures of illness severity are rarely recorded for common infection consultations in primary care, so it was not possible to adjust for illness severity in analyses. It is also established that not all infection consultations in primary care are correctly coded, leading to underestimation of consultation rates [7]. We included data from 706 general practices over a 16-year period. Infection consultation and antibiotic prescribing rates were estimated from sample data. The estimates in this paper represent average values for this population of general practices and period of time. However, we conducted a sensitivity analysis with data from 2014 to 2017 only. We also acknowledge that in addition to changes in overall antibiotic utilisation, there have been changes in the proportion of prescriptions for broad-spectrum antibiotics. Future studies might be designed to compare the probability of sepsis if broad-spectrum or narrow-spectrum antibiotics are prescribed. The sample design used to estimate infection consultation rates and antibiotic prescribing proportions gave each practice, and each study year, equal weight, but we could have weighted the sample by practice size.

We analysed data for infection consultations in primary care and compared outcomes if antibiotics were or were not prescribed. However, previous studies showed that antibiotics may be prescribed at consultations with no definite diagnosis recorded [7,25]. We did not include these prescriptions because there was no valid comparator in terms of consultations without antibiotic prescriptions, but conclusions might have differed if missing diagnosis information had been complete. We caution that the precise values of these estimates may be expected to vary in different local contexts and according to the types of infection circulating in a community at a given time. We did not employ the approach of null hypothesis significance testing and do not report P values. We evaluated association modification by age, gender, frailty level, and consultation type. We employed the e-Frailty Index, which is a well-described measure based on 36 deficits [26], although we also applied it in the age range of 55–64 years, in which it is less well documented. We estimated stratified values for broad groups of patients, defined in terms of age, gender, and frailty. We did not estimate personalised risks for individual patients, and the clinical circumstances in each specific consultation should be used to inform estimates of sepsis risk for individuals. We relied on clinical records of sepsis from general practice, but we cannot be sure that all sepsis events were community rather than hospital acquired. In the UK, patients register with a family practice for continuing care, but patients may utilise emergency and out-of-hours services for acute problems such as sepsis, and these events might not be captured in primary care records. Providers may vary in their use of the term ‘sepsis’, which is an intermediate condition linking an infection and organ damage consequent on infection. The selection of clinical terms and medical codes is at the discretion of clinical staff, leading to lack of data standardisation. The conditions identified as ‘sepsis’ may represent a range of disease severity, and probability estimates might be proportionately lower if only severe sepsis was included. However, by using linked data, we showed that inclusion of hospital episodes and mortality records did not lead to any important changes in conclusions. Further research is needed to refine, update, and improve the accuracy of these initial estimates.

Comparison with other studies

There has been a trend toward more-frequent recording of sepsis in recent years, and this has not always been accompanied by evidence of increased blood stream infections. In an interrupted time series analysis, Balinskaite and colleagues [31] found no evidence that antimicrobial stewardship interventions in the UK might be associated with increased rates of sepsis. In an ecological analysis [24], we did not find evidence that general practices with lower antibiotic prescribing might have greater risk of sepsis over the same period of time and in the same practices as were included in the present study. Gharbi and colleagues [32] found that in older adults presenting with UTI, there was increased risk of sepsis if antibiotic prescriptions were not given or were delayed. The present results extend these findings by estimating risks across all ages, different levels of frailty, and different types of infection consultations. The lack of consistency between estimates from ecological- and individual-level analyses is likely to be explained by the substantial proportion of patients with sepsis who present without preceding infection consultations in primary care, as well as the small proportion of higher-risk consultations that are not associated with antibiotic prescriptions. RTI consultations are extremely frequent, which may account for the lower probability of associated sepsis. Respiratory infections are often the result of virus infections, and clinicians may tend to reserve the term ‘sepsis’ for bacterial infections. We evaluated uncomplicated lower UTIs, but estimates for the probability of sepsis might have been higher if kidney infections had been included.

Conclusions

This paper quantifies the risk of sepsis following common infection consultations in primary care. These may be used in antimicrobial stewardship to identify groups of consultations at which reduction of antibiotic prescribing can be pursued more safely. The estimates show that risks of sepsis and benefits of antibiotics are generally more substantial among older adults, persons with more advanced frailty, or following UTI.

Supporting information

S1 STROBE Checklist. Items that should be included in reports of cohort studies.

STROBE, Strengthening the Reporting of Observational Studies in Epidemiology.

(DOC)

Click here for additional data file.^{(99KB, doc)}

S1 Table. List of Read codes for sepsis.

(XLSX)

Click here for additional data file.^{(13.8KB, xlsx)}

S2 Table. List of Read codes for common infections.

(XLSX)

Click here for additional data file.^{(38.9KB, xlsx)}

S3 Table. List of product codes for antibiotics.

(XLSX)

Click here for additional data file.^{(78.2KB, xlsx)}

S4 Table. Proportion of consultations with antibiotics prescribed and consultation rates per person-year for different common infections.

(DOCX)

Click here for additional data file.^{(31.3KB, docx)}

S5 Table. Estimated distribution of CPRD GOLD population by frailty level.

CPRD, Clinical Practice Research Datalink; PY, sum of person-years from 2002 to 2017.

(DOCX)

Click here for additional data file.^{(29.3KB, docx)}

S6 Table. Distribution of sepsis cases by gender, region, and period.

(DOCX)

Click here for additional data file.^{(27.1KB, docx)}

S7 Table. Estimates by frailty category.

(DOCX)

Click here for additional data file.^{(35.2KB, docx)}

S8 Table. Estimates by type of infection consultation.

(DOCX)

Click here for additional data file.^{(42.7KB, docx)}

S9 Table. Sensitivity analysis using data for 2014–2017 only.

Column headings as main text Table 2.

(DOCX)

Click here for additional data file.^{(29.9KB, docx)}

S1 Fig. Flow chart showing participant selection for main and linked samples.

(DOCX)

Click here for additional data file.^{(47.4KB, docx)}

S2 Fig. Estimates for number of antibiotic prescriptions needed to prevent one sepsis episode (NNT) for four periods: 2002–2005 (blue), 2006–2009 (green), 2010–2013 (orange), and 2014–2017 (red).

NNT, number needed to treat.

(DOCX)

Click here for additional data file.^{(44.2KB, docx)}

S3 Fig. Probability of an infection consultation in primary care in the 30 days preceding a sepsis diagnosis using CPRD (linked sample) records (red); CPRD and linked HES records (blue); or CPRD, HES, and linked ONS mortality records (green).

CPRD, Clinical Practice Research Datalink; HES, Hospital Episode Statistics; ONS, Office for National Statistics.

(DOCX)

Click here for additional data file.^{(45.7KB, docx)}

S4 Fig. Estimated probability (95% uncertainty interval) of a first sepsis event within 30 days of an infection consultation in primary care if antibiotics were prescribed.

CPRD (linked sample) records only (red); CPRD and linked HES records (blue); or CPRD, HES, and linked ONS mortality records (green). CPRD, Clinical Practice Research Datalink; HES, Hospital Episode Statistics; ONS, Office for National Statistics.

(DOCX)

Click here for additional data file.^{(43.3KB, docx)}

S5 Fig. Estimated number of antibiotic prescriptions (95% uncertainty interval) to prevent a first sepsis event within 30 days of an infection consultation in primary care.

(DOCX)

Click here for additional data file.^{(48.4KB, docx)}

Acknowledgments

The SafeABStudy Group also includes Dr Olga Boiko, Dr Caroline Burgess, Dr Vasa Curcin, and Dr James Shearer.

The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health. The authors had full access to all the data in the study, and all authors shared final responsibility for the decision to submit for publication.

Abbreviations

AMR: antimicrobial drug resistance
APC: admitted patient care;CPRD, Clinical Practice Research Datalink
HES: Hospital Episode Statistics
ISAC: Independent Scientific Advisory Committee
NEWS2: National Early Warning Score
NNT: number needed to treat
ONS: Office for National Statistics
RTI: respiratory tract infection
STROBE: Strengthening the Reporting of Observational Studies in Epidemiology
UI: uncertainty interval
UTI: urinary tract infection.

Data Availability

Data cannot be shared publicly because they are analysed under licence. Permission for data access is through the CPRD Independent Scientific Advisory Committee (ISAC, contact via isac@mhra.gov.uk) for researchers who meet the criteria for access to confidential data. The data underlying the results presented in the study are available from the Clinical Practice Research Datalink (CPRD, cprdenquiries@mhra.gov.uk).

Funding Statement

The study is funded by the National Institute for Health Research (NIHR) Health Services and Delivery Programme (16-116-46). The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

References

1.Chan M. WHO Director-General addresses ministerial conference on antimicrobial resistance. Geneva: World Health Organization; 2016. [cited 2020 Jul 8]. Available from: https://www.who.int/dg/speeches/2016/antimicrobial-resistance-conference/en/. [Google Scholar]
2.Public Health England. English Surveillance Programme for Antimicrobial Utilisation and Resistance (ESPAUR). Report 2018–2019. London: Public Health England; 2019. [cited 2020 Jul 8]. Available from: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/843129/English_Surveillance_Programme_for_Antimicrobial_Utilisation_and_Resistance_2019.pdf. [Google Scholar]
3.Gulliford MC, Dregan A, Moore MV, Ashworth M, Staa T, McCann G, et al. Continued high rates of antibiotic prescribing to adults with respiratory tract infection: survey of 568 UK general practices. BMJ Open. 2014;4:e006245 10.1136/bmjopen-2014-006245 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Ashworth M, Latinovic R, Charlton J, Cox K, Rowlands G, Gulliford M. Why has antibiotic prescribing for respiratory illness declined in primary care? A longitudinal study using the General Practice Research Database. J Public Health (Oxf). 2004;26:268–74. [DOI] [PubMed] [Google Scholar]
5.Gulliford M, Latinovic R, Charlton J, Little P, van Staa T, Ashworth M. Selective decrease in consultations and antibiotic prescribing for acute respiratory tract infections in UK primary care up to 2006. J Public Health (Oxf). 2009;31:512–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Aabenhus R, Hansen MP, Siersma V, Bjerrum L. Clinical indications for antibiotic use in Danish general practice: results from a nationwide electronic prescription database. Scand J Prim Health Care. 2017;35:162–9. 10.1080/02813432.2017.1333321 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Dolk FCK, Pouwels KB, Smith DRM, Robotham JV, Smieszek T. Antibiotics in primary care in England: which antibiotics are prescribed and for which conditions? J Antimicrobial Chemotherapy. 2018;73(suppl_2):ii2–ii10. 10.1093/jac/dkx504 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Department of Health. UK 5 Year Antimicrobial Resistance Strategy 2013 to 2018. London: Department of Health; 2013. [Google Scholar]
9.The Review on Antimicrobial Resistance by Jim O’Neill. Tackling drug-resistant infections globally: final report and recommendations. London: Review on antimicrobial resistance; 2016. [cited 2020 Jul 8]. Available from: https://amr-review.org/. [Google Scholar]
10.Gulliford MC, Moore MV, Little P, Hay AD, Fox R, Prevost AT, et al. Safety of reduced antibiotic prescribing for self limiting respiratory tract infections in primary care: cohort study using electronic health records. BMJ. 2016;354:i3410 10.1136/bmj.i3410 [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Gulliford MC, Juszczyk D, Prevost AT, Soames J, McDermott L, Sultana K, et al. Electronically delivered interventions to reduce antibiotic prescribing for respiratory infections in primary care: cluster RCT using electronic health records and cohort study. Health Technol Assess. 2019;23(11):1–70. 10.3310/hta23110 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Cabral C, Lucas PJ, Ingram J, Hay AD, Horwood J. “It's safer to …” parent consulting and clinician antibiotic prescribing decisions for children with respiratory tract infections: An analysis across four qualitative studies. Soc Sci Med. 2015. [cited 2020 Jul 8];136–137:156–64. Available from: 10.1016/j.socscimed.2015.05.027. [DOI] [PubMed] [Google Scholar]
13.Lucas PJ, Cabral C, Hay AD, Horwood J. A systematic review of parent and clinician views and perceptions that influence prescribing decisions in relation to acute childhood infections in primary care. Scand J Prim Health Care. 2015;33:11–20. 10.3109/02813432.2015.1001942 [DOI] [PMC free article] [PubMed] [Google Scholar]
14.van den Broek d'Obrenan J, Verheij TJM, Numans ME, van der Velden AW. Antibiotic use in Dutch primary care: relation between diagnosis, consultation and treatment. J Antimicrobial Chemotherapy. 2014;69:1701–7. 10.1093/jac/dku005 [DOI] [PubMed] [Google Scholar]
15.Lusini G, Lapi F, Sara B, Vannacci A, Mugelli A, Kragstrup J, et al. Antibiotic prescribing in paediatric populations: a comparison between Viareggio, Italy and Funen, Denmark. Eur J Public Health. 2009;19:434–8. 10.1093/eurpub/ckp040 [DOI] [PubMed] [Google Scholar]
16.Laxminarayan R, Matsoso P, Pant S, Brower C, Røttingen J-A, Klugman K, et al. Access to effective antimicrobials: a worldwide challenge. The Lancet. 2016. [cited 2020 Jul 8];387:168–75. Available from: 10.1016/S0140-6736(15)00474-2. [DOI] [PubMed] [Google Scholar]
17.NHS England. Quality Premium: 2016/17 Guidance for CCGs. Leeds: NHS England; 2016 [cited 2020 Jun 1]. Available from: https://www.england.nhs.uk/wp-content/uploads/2016/03/qualty-prem-guid-2016-17.pdf.
18.NHS Choices. Sepsis. London: NHS; 2020 [cited 2020 Jul 8]. Available from: http://www.nhs.uk/Conditions/Blood-poisoning/Pages/Introduction.aspx.
19.National Institute for Health and Care Excellence. Sepsis: recognition, diagnosis and early management. NICE guideline [NG51]. London: NICE; 2017. Available from: https://www.nice.org.uk/guidance/ng51. [PubMed]
20.Royal College of Physicians. National Early Warning Score (NEWS) 2. London: Royal College of Physicians; 2017 [cited 2020 Jul 8]. Available from: https://www.rcplondon.ac.uk/projects/outputs/national-early-warning-score-news-2.
21.The Lancet Respiratory Medicine. Crying wolf: the growing fatigue around sepsis alerts. Lancet Respiratory Medicine. 2018;6:161 10.1016/S2213-2600(18)30072-9 [DOI] [PubMed] [Google Scholar]
22.Herrett E, Gallagher AM, Bhaskaran K, Forbes H, Mathur R, van Staa T, et al. Data Resource Profile: Clinical Practice Research Datalink (CPRD). Int J Epidemiol. 2015;44:827–36. 10.1093/ije/dyv098 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Herrett E, Thomas SL, Schoonen WM, Smeeth L, Hall AJ. Validation and validity of diagnoses in the General Practice Research Database: a systematic review. Br J Clin Pharmacol. 2010;69:4–14. 10.1111/j.1365-2125.2009.03537.x [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Gulliford MC, Sun X, Charlton J, Winter JR, Bunce C, Boiko O et al. Serious bacterial infections and antibiotic prescribing in primary care. Cohort study using electronic health records in the UK. BMJ open. 2020;10:e036975 10.1136/bmjopen-2020-036975 [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Sun X, Gulliford M. Reducing antibiotic prescribing in primary care in England from 2014 to 2017: Population-based cohort study. BMJOpen. 2019;9:e023989 10.1136/bmjopen-2018-023989 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Clegg A, Bates C, Young J, Ryan R, Nichols L, Ann Teale E, et al. Development and validation of an electronic frailty index using routine primary care electronic health record data. Age and Ageing. 2016;45:353–60. 10.1093/ageing/afw039 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Gafoor R, Charlton J, Ravindrarajah R, Gulliford MC. Importance of Frailty for Association of Antipsychotic Drug Use with Risk of Fracture: Cohort Study Using Electronic Health Records. J Am Med Dir Assoc. 2019;20:1495–1501.e1. 10.1016/j.jamda.2019.05.009 [DOI] [PubMed] [Google Scholar]
28.Winter J, Charlton J, Ashworth M, Bunce C, Gulliford MC. Peritonsillar abscess and antibiotic prescribing for respiratory infection in primary care. Population-based cohort study and decision analytic model. Ann Fam Med. Forthcoming 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Gelman A, Carlin JB, Stern HS, Rubin DB. Bayesian Data Analysis. 3rd ed. Boca Raton, FL: Chapman and Hall / CRC; 2013. [Google Scholar]
30.Rhee C, Murphy MV, Li L, Platt R, Klompas M. Comparison of trends in sepsis incidence and coding using administrative claims versus objective clinical data. Clin Inf Dis 2015;60:88–95. 10.1093/cid/ciu750 [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Balinskaite V, Bou-Antoun S, Johnson AP, Holmes A, Aylin P. An Assessment of Potential Unintended Consequences Following a National Antimicrobial Stewardship Program in England: An Interrupted Time Series Analysis. Clinical Infectious Diseases. 2019;69:233–42. 10.1093/cid/ciy904 [DOI] [PubMed] [Google Scholar]
32.Gharbi M, Drysdale JH, Lishman H, Goudie R, Molokhia M, Johnson AP, et al. Antibiotic management of urinary tract infection in elderly patients in primary care and its association with bloodstream infections and all-cause mortality: population-based cohort study. BMJ. 2019;364:l525 10.1136/bmj.l525 [DOI] [PMC free article] [PubMed] [Google Scholar]

PLoS Med. doi: 10.1371/journal.pmed.1003202.r001

Decision Letter 0

Artur Arikainen

14 Apr 2020

Dear Dr Gulliford,

Thank you for submitting your manuscript entitled "PROBABILITY OF SEPSIS AFTER INFECTION CONSULTATIONS IN PRIMARY CARE. Population based cohort study and decision analytic model" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by .

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Artur Arikainen,

Associate Editor

PLOS Medicine

PLoS Med. doi: 10.1371/journal.pmed.1003202.r002

Decision Letter 1

Artur Arikainen

30 Apr 2020

Dear Dr. Gulliford,

Thank you very much for submitting your manuscript "PROBABILITY OF SEPSIS AFTER INFECTION CONSULTATIONS IN PRIMARY CARE. Population based cohort study and decision analytic model" (PMEDICINE-D-20-01208R1) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by May 21 2020 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/plosmedicine/s/submission-guidelines#loc-methods.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Artur Arikainen,

Associate Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

1. We ask that you address all points raised by the 3 reviewers, but want to draw particular attention to the following: Please address the reviewers’ requests regarding better reporting of your methodology. In particular, we ask that you provide a detailed response to the comments of reviewer #3 on the suitability of the dataset for your analyses. Please also consider the comments of reviewer #2 on whether it would be possible to include analyses by antibiotic class, though we recognise that this may not be feasible at this stage.

2. Please revise your title according to PLOS Medicine's style, using sentence case and a colon before the study type. Please also mention that the study is based on data from the UK.

3. Please structure your abstract using the PLOS Medicine headings (Background, Methods and Findings, Conclusions). Please combine the Design, Setting, Participants etc. sections into one section: “Methods and findings”. In the last sentence of the Methods and Findings section, please describe the main limitation(s) of the study's methodology. Please also include summary demographic information (eg. age, sex) for the cohort as a whole.

4. Please use the term ‘patient’ rather than ‘participant’ throughout your manuscript.

5. The Data Availability Statement (DAS) requires revision. For each data source used in your study: If the data are owned by a third party but available upon request, please note this and state the owner of the data set and contact information for data requests (web or email address). Note that a study author cannot be the contact person for the data. If the data are not freely available, please describe briefly the ethical, legal, or contractual restriction that prevents you from sharing it.

6. At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

7. Please insert your citations before punctuation marks, eg: “…is attracting the concern of national governments and international organisations [1].” Please also ensure that your reference list is presented in the Vancouver style, including bold and italic formatting, where appropriate.

8. Please include a full list of the read codes used in your study, as Supporting Information.

9. Please provide line numbers throughout your manuscript text.

10. In the Abstract and Main Text, please include significance p values for the numerical data presented.

11. Please provide a table showing the baseline characteristics of the study population.

12. Please ensure that the study is reported according to the STROBE guidelines (https://www.strobe-statement.org/index.php?id=available-checklists), and include the completed STROBE checklist as Supporting Information. When completing the checklist, please use section and paragraph numbers, rather than page numbers. Please add the following statement, or similar, to the Methods: "This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist).”

13. Regarding reference [28] listed as ‘in press’, papers cannot be listed in the reference list until they have been accepted for publication or are otherwise publicly accessible (for example, in a preprint archive). The information may be cited in the text as a personal communication with the author if the author provides written permission to be named. Alternatively, please provide a different appropriate reference.

14. Where possible, please provide additional geographic information relating to your dataset: for instance, are all practices from a small number of cities or is it an even distribution through the UK?

15. You can omit the Data, Funding, Conflict of Interest, and Author Contributions information on page 14, as this is extracted from the data provided in the submission form.

----

Comments from the reviewers:

Reviewer #1: "PROBABILITY OF SEPSIS AFTER INFECTION CONSULTATIONS IN PRIMARY CARE. Population based cohort study and decision analytic model" attempts to estimate the contribution of antibiotic prescription towards the prevention of sepsis, on over 66 million person-years of data collected from 706 U.K. general practices, from 2002 to 2017.

The major motivating concern is a trade-off between overprescription of antibiotics (wastage of resources, possible development of resistance by bacteria) and underprescription (reduces immediate patient safety, by increasing risk of follow-up bacteria infections, e.g. sepsis). This work focuses on the latter, and reports the estimated number of antibiotic prescriptions required to prevent one episode of sepsis (NNT) by age stratification. The extremely low incidence of sepsis in the month following an infection consultation (<0.0005, for all age groups) implies the need for an appropriately large set of data for correlations to be reliably drawn, a requirement that seems to be fulfilled. These NNTs were computed following a decision tree model, based on Bayes' theorem (Figure 1).

It was observed that NNT is always positive (i.e. antibiotic prescription is always correlated with reduced sepsis), and that NNT generally decreases with age (i.e. the contribution of antibiotic prescription towards reducing sepsis, increases with age). The relatively large values for NNT (minimum 262, maximum 65522) appear to mostly be a consequence of the extremely low incidence of follow-up sepsis, i.e. P(Sepsis|Infection), to begin with. The major limitation of this model, that the selection of patients receiving antibiotics is likely not random (i.e. P(AB|Infection) being likely dependant on other non-considered factors), was discussed; in any case, this likely leans towards NNT being overestimated, i.e. the impact of antibiotics on sepsis prevention being understated, from the presented results.

A particular strength of this study is its employment of a diverse set of records collected from hundreds of practices, covering tens of millions of patient-years, which allowed for a retrospective analysis of the antibiotics-sepsis relationship that would otherwise have been impractical through trials. The main decision tree model is clearly structured and described, appropriate sensitivity analyses were considered, and the conclusions appear broadly valid and of interest.

There do remain a number of possible clarifications:

1. In the "Sepsis events" section (Page 5), it is stated that "...we identified consultations for respiratory tract infections, skin infections and urinary tract infections because these are the most important groups of conditions for which antibiotics are prescribed in primary care". This however seems to imply that sepsis consultations for which antibiotic prescriptions were prescribed in the past month (i.e. for infections that weren't respiratory tract/skin/urinary tract), might have been ignored. While it is stated that these are the "most important group" of conditions for which antibiotics are prescribed, it is not clear whether they are also the most frequent group. In other words, was there a significant quantity of antibiotics prescriptions that were not covered under these three infection categories, and thus not analyzed? The supplementary data of the cited reference [25] was examined, but did not appear to cover this detail either.

2. In the "Selection of sample for antibiotic prescribing analysis" section, it is stated that a random sample of patients was drawn by selecting 10 participants for each gender/age group, also stratified by family practice.

a) From our understanding, the model probabilities presented in Table 1 were obtained from this set of sampled patients. If so, this might be explicitly stated.

b) It is not obvious as to why the stratification by practices was required. This seems to imply that, for example, a small practice with exactly ten female patients in the 0-4 year age group for some year would have all ten of them sampled for analysis, while a large practice with say 300 such patients would then also have just 10 of them sampled, with the remaining 290 ignored. If this description is correct, while this might reduce certain geographic-based biases, it would nevertheless seem to omit much entirely-valid data from consideration. Moreover, even if normalization by individual family practices is desired, another option might be to compute the required probabilities with all valid patients at the practice level, and then aggregate these probabilities equally, rather than sample (and discard data from consideration) early on.

c) Following from the above, it might be informative to have a flowchart collating the state of the data pre- and post-sampling (e.g. from the pre-sampled 66.2 million person-years to how many person-years post-sampling, from how many individual participants pre-sampling to 671,830 participants post-sampling, etc)

3. The degree of comprehensiveness of the EHR data might be discussed, i.e. might an individual patient visit one family practice for an infection, then visit a different practice/hospital upon onset of sepsis (not necessarily through referral)? If this is possible, would such cases be recognized/considered in the records/analysis? Also, given the relatively high mortality for sepsis (59,000 deaths from somewhat over 200,000 hospital admissions, from the Introduction), would deaths from sepsis (and possibly other complications) be expected to be recognized in the relevant records?

4. It is not discussed as to why infection/antibiotics prescription within the preceding 30 days was determined as the relavant period, rather than e.g. 60 days. Is 30 days a standard assumption for duration of antibiotic effect?

Reviewer #2: The authors sampled the CPRD GOLD primary care electronic health record data base to estimate the probability of consultation for an infection of the skin, respiratory tract or urinary tract, and used a decision tree approach to estimate the probability of sepsis following these consultations by whether an antibiotic was prescribed. Strengths of the study include population estimates by age, sex and frailty of this risk overall and by each infection, and provide estimates of the number of antibiotic prescriptions required to prevent one sepsis event for each of the above mentioned categories. The study uses a strong observational design and sampling method.

There are a number of areas where the presentation and interpretation of might strengthened. First, the outcome was determined by ICD codes (presumably assigned by treating physicians or administrative staff), so there is no uniform definition of the different categories. This should be acknowledged in the limitations. Second, sepsis, urosepsis and septicemia are considered to be equivalent and are lumped together in the presentation. These conditions may vary significantly in severity, and if sensitive to the prescribed antibiotic treatment can be straightforward (septicemia and urosepsis). At a minimum, it would be very helpful to provide separate estimates for sepsis.

The two most common sources for sepsis are the lungs (pneumonia) and the kidneys. However, respiratory illnesses (including pneumonia) are often caused by viruses, although the viral infection may lead to secondary bacterial infection (e.g. influenza). Therefore, it is not surprising that the NNT is highest for respiratory infections. If there were a way to disentangle this group (perhaps presenting results for pneumonia, alone), would be useful. Similarly, it would be useful to separate UTI into cystitis and pyelonephritis.

Discussing how the authors believe their data might help antibiotics might be used more safely in greater detail would strengthen the manuscript. In this regard, it is unfortunate that the authors did not examine classes of antibiotics prescribed. One strategy for reducing emergence of antibiotic resistance is to minimize use of broad-spectrum antibiotics where possible. With additional analyses, the authors might speak more directly to this issue.

The justification for choosing sepsis as an outcome might be made more explicitly in the introduction and explored more thoroughly in the discussion. Appropriate antibiotic therapy has an immediate benefit that is not tied to the risk of sepsis. Sadly, the only way to limit the emergence of antibiotic resistance - which threatened our ability to treat severe infection such as sepsis - is to limit antibiotic use. It may not be reasonable to treat 30,000 individuals to prevent one case of sepsis, but there may be other reasons to do so. On the flip side - depending on the antibiotic - treating 30,000 individuals might results in 300 adverse side effects.

Reviewer #3: The findings of this study could have important clinical implications for practitioners that fear missing sepsis. Expressing findings in NNT is very helpful and will support understanding the clinical relevance of the findings for frontline practitioners.

My major concern with this study concerns the reliability of these estimates given they have drawn upon data from such a large period of time. The researchers are / were up against at least three major challenges: 1. Sepsis is a tricky subject for carrying out data-linkage studies because definitions have changed multiple times over the last couple of years alone, the awareness amongst both clinicians and the public has shifted dramatically, and there have been several preventive health interventions aiming to tackle delays in the identification and management of sepsis. 2. Depending on the license agreement with CPRD, there can be limits on the number of records extracted. 3. Best practice, particularly concerning antibiotic stewardship, has changed considerably since 2002 (18 years ago). To mitigate the influence of these challenges, I am perplexed they did not endeavour to sample more patients (to the maximum of their license) from more recent years (even 2012-2017), or at the very least in the manuscript explored this possibility +/- considered the implications of not doing so (they might have, and not had sufficient word count to do so here). The horse has bolted now and they can't fix this.

Any attachments provided with reviews can be seen via the following link:

[LINK]

PLoS Med. 2020 Jul 23;17(7):e1003202. doi: 10.1371/journal.pmed.1003202.r003

Author response to Decision Letter 1

18 May 2020

Attachment

Submitted filename: ResponseToReviewers18May2020.docx

Click here for additional data file.^{(36.9KB, docx)}

PLoS Med. doi: 10.1371/journal.pmed.1003202.r004

Decision Letter 2

Artur Arikainen

28 May 2020

Dear Dr. Gulliford,

Thank you very much for re-submitting your manuscript "Probability of sepsis after infection consultations in primary care in the United Kingdom: population-based cohort study and decision analytic model" (PMEDICINE-D-20-01208R2) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by two reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

Our publications team (plosmedicine@plos.org) will be in touch shortly about the production requirements for your paper, and the link and deadline for resubmission. DO NOT RESUBMIT BEFORE YOU'VE RECEIVED THE PRODUCTION REQUIREMENTS.

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Jun 04 2020 11:59PM.

Sincerely,

Artur Arikainen,

Associate Editor

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

1. Please update the title to include the study dates: “Probability of sepsis after infection consultations in primary care in the United Kingdom in 2002-17: population-based cohort study and decision analytic model”

2. Please update your Competing Interests statement on the submission form to the following standard text: “The authors have declared that no competing interests exist.”

3. Please move the “Data sources” section from page 18 to either the Data Availability Statement in the submission form, or the Methods section of the main text, or remove it altogether.

4. In the Abstract, please include an additional limitation, eg. the possibility of missing or incorrect health record data, or possible sources of antibiotics outside primary care.

5. Please remove the keywords from page 2. Our published articles are indexed automatically using a controlled taxonomy.

6. Author summary: Please spell out UTI and RTI, for clarity to non-scientist readers.

7. Please include line numbers in your manuscript margin.

8. In the section “Data source”, please provide a URL link to the database website.

9. Please cite the study protocol the same way as with other references, rather than as a hyperlink, or include the URL in brackets.

10. In the section “Selection of sample for antibiotic prescribing analysis”, please include a brief description of how the random sample was chosen, eg. by computer-generated list.

11. There are some instances in the results where UTI and RTI are spelled out, even though the abbreviations are already used in earlier parts of the text, eg. page 12.

12. In the Discussion, please break up the long paragraph on limitations, in order to improve readability.

13. Thank you for addressing our comment relating to p values. Our only request is that you remove this sentence: “Readers may reflect on the substantive importance of estimated differences, and associated uncertainty intervals, for their work.”

14. Please correct this sentence in the Discussion to: “Future studies might be designed to compare the probability of sepsis if broad-spectrum or narrow-spectrum antibiotics are prescribed.”

15. Please format your references to strict Vancouver style – bold and italics are not used.

16. Please correct the typo in reference 9: “Antimicrobial”

17. Please provide more access details (eg. a URL) for references 17, 19, and please update reference 28 to include full details rather than “in press”.

18. In the Discussion please replace ‘significant’ in the following sentence with a more appropriate term, eg. ‘notable’: “The lack of consistency between estimates from ecological- and individual-level analyses are likely to be explained by the significant proportion of patients…”

19. Please avoid the use of ‘effect’ throughout your text, given the observational nature of your study, eg. as in this sentence: “Age, gender and frailty were evaluated as effect modifiers.”

20. The terms gender and sex are not interchangeable (as discussed in http://www.who.int/gender/whatisgender/en/); please use the appropriate term.

21. Thank you for responding to our comment 14 in the previous decision letter. To clarify, where possible, we would like you to provide a summary of sepsis events broken down by region or NHS trust, eg. as Supporting Information.

-------

Comments from Reviewers:

Reviewer #1: We thank the authors for considering our previous suggestions. For Supplementary Figure 1, the arrows might be labelled with brief descriptions of the selection process for convenience, or the sampling description summarized as a caption. On the additional sensitivity analyses for 2002-2005 & 2014-2017, the authors might consider including the intervening four year periods (2006-2009, 2010-2013) as well for completeness, if it is not too much trouble.

Reviewer #3: My comments have been adequately investigated and now addressed in the manuscript with appropriate discussion of their implications. The sensitivity analyses highlight the fragility of these data when different assumptions are taken. The authors have sufficiently described such limitations in the main paper, and should consider reflecting this more explicitly in the abstract.

Any attachments provided with reviews can be seen via the following link:

[LINK]

PLoS Med. 2020 Jul 23;17(7):e1003202. doi: 10.1371/journal.pmed.1003202.r005

Author response to Decision Letter 2

23 Jun 2020

Attachment

Submitted filename: ResponseToReviews12June2020.docx

Click here for additional data file.^{(32.5KB, docx)}

PLoS Med. doi: 10.1371/journal.pmed.1003202.r006

Decision Letter 3

Artur Arikainen

25 Jun 2020

Dear Prof. Gulliford,

On behalf of my colleagues and the academic editor, Dr. Andrew Carson-Stevens, I am delighted to inform you that your manuscript entitled "Probability of sepsis after infection consultations in primary care in the United Kingdom in 2002-17: population-based cohort study and decision analytic model" (PMEDICINE-D-20-01208R3) has been accepted for publication in PLOS Medicine.

PRODUCTION PROCESS

Before publication you will see the copyedited word document (in around 1-2 weeks from now) and a PDF galley proof shortly after that. The copyeditor will be in touch shortly before sending you the copyedited Word document. We will make some revisions at the copyediting stage to conform to our general style, and for clarification. When you receive this version you should check and revise it very carefully, including figures, tables, references, and supporting information, because corrections at the next stage (proofs) will be strictly limited to (1) errors in author names or affiliations, (2) errors of scientific fact that would cause misunderstandings to readers, and (3) printer's (introduced) errors.

If you are likely to be away when either this document or the proof is sent, please ensure we have contact information of a second person, as we will need you to respond quickly at each point.

PRESS

A selection of our articles each week are press released by the journal. You will be contacted nearer the time if we are press releasing your article in order to approve the content and check the contact information for journalists is correct. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact.

PROFILE INFORMATION

Now that your manuscript has been accepted, please log into EM and update your profile. Go to https://www.editorialmanager.com/pmedicine, log in, and click on the "Update My Information" link at the top of the page. Please update your user information to ensure an efficient production and billing process.

Thank you again for submitting the manuscript to PLOS Medicine. We look forward to publishing it.

Best wishes,

Artur Arikainen,

Associate Editor

PLOS Medicine

plosmedicine.org

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 STROBE Checklist. Items that should be included in reports of cohort studies.

STROBE, Strengthening the Reporting of Observational Studies in Epidemiology.

(DOC)

Click here for additional data file.^{(99KB, doc)}

S1 Table. List of Read codes for sepsis.

(XLSX)

Click here for additional data file.^{(13.8KB, xlsx)}

S2 Table. List of Read codes for common infections.

(XLSX)

Click here for additional data file.^{(38.9KB, xlsx)}

S3 Table. List of product codes for antibiotics.

(XLSX)

Click here for additional data file.^{(78.2KB, xlsx)}

S4 Table. Proportion of consultations with antibiotics prescribed and consultation rates per person-year for different common infections.

(DOCX)

Click here for additional data file.^{(31.3KB, docx)}

S5 Table. Estimated distribution of CPRD GOLD population by frailty level.

CPRD, Clinical Practice Research Datalink; PY, sum of person-years from 2002 to 2017.

(DOCX)

Click here for additional data file.^{(29.3KB, docx)}

S6 Table. Distribution of sepsis cases by gender, region, and period.

(DOCX)

Click here for additional data file.^{(27.1KB, docx)}

S7 Table. Estimates by frailty category.

(DOCX)

Click here for additional data file.^{(35.2KB, docx)}

S8 Table. Estimates by type of infection consultation.

(DOCX)

Click here for additional data file.^{(42.7KB, docx)}

S9 Table. Sensitivity analysis using data for 2014–2017 only.

Column headings as main text Table 2.

(DOCX)

Click here for additional data file.^{(29.9KB, docx)}

S1 Fig. Flow chart showing participant selection for main and linked samples.

(DOCX)

Click here for additional data file.^{(47.4KB, docx)}

NNT, number needed to treat.

(DOCX)

Click here for additional data file.^{(44.2KB, docx)}

CPRD, Clinical Practice Research Datalink; HES, Hospital Episode Statistics; ONS, Office for National Statistics.

(DOCX)

Click here for additional data file.^{(45.7KB, docx)}

S4 Fig. Estimated probability (95% uncertainty interval) of a first sepsis event within 30 days of an infection consultation in primary care if antibiotics were prescribed.

(DOCX)

Click here for additional data file.^{(43.3KB, docx)}

S5 Fig. Estimated number of antibiotic prescriptions (95% uncertainty interval) to prevent a first sepsis event within 30 days of an infection consultation in primary care.

(DOCX)

Click here for additional data file.^{(48.4KB, docx)}

Attachment

Submitted filename: ResponseToReviewers18May2020.docx

Click here for additional data file.^{(36.9KB, docx)}

Attachment

Submitted filename: ResponseToReviews12June2020.docx

Click here for additional data file.^{(32.5KB, docx)}

Data Availability Statement

[pmed.1003202.ref001] 1.Chan M. WHO Director-General addresses ministerial conference on antimicrobial resistance. Geneva: World Health Organization; 2016. [cited 2020 Jul 8]. Available from: https://www.who.int/dg/speeches/2016/antimicrobial-resistance-conference/en/. [Google Scholar]

[pmed.1003202.ref002] 2.Public Health England. English Surveillance Programme for Antimicrobial Utilisation and Resistance (ESPAUR). Report 2018–2019. London: Public Health England; 2019. [cited 2020 Jul 8]. Available from: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/843129/English_Surveillance_Programme_for_Antimicrobial_Utilisation_and_Resistance_2019.pdf. [Google Scholar]

[pmed.1003202.ref003] 3.Gulliford MC, Dregan A, Moore MV, Ashworth M, Staa T, McCann G, et al. Continued high rates of antibiotic prescribing to adults with respiratory tract infection: survey of 568 UK general practices. BMJ Open. 2014;4:e006245 10.1136/bmjopen-2014-006245 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref004] 4.Ashworth M, Latinovic R, Charlton J, Cox K, Rowlands G, Gulliford M. Why has antibiotic prescribing for respiratory illness declined in primary care? A longitudinal study using the General Practice Research Database. J Public Health (Oxf). 2004;26:268–74. [DOI] [PubMed] [Google Scholar]

[pmed.1003202.ref005] 5.Gulliford M, Latinovic R, Charlton J, Little P, van Staa T, Ashworth M. Selective decrease in consultations and antibiotic prescribing for acute respiratory tract infections in UK primary care up to 2006. J Public Health (Oxf). 2009;31:512–20. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref006] 6.Aabenhus R, Hansen MP, Siersma V, Bjerrum L. Clinical indications for antibiotic use in Danish general practice: results from a nationwide electronic prescription database. Scand J Prim Health Care. 2017;35:162–9. 10.1080/02813432.2017.1333321 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref007] 7.Dolk FCK, Pouwels KB, Smith DRM, Robotham JV, Smieszek T. Antibiotics in primary care in England: which antibiotics are prescribed and for which conditions? J Antimicrobial Chemotherapy. 2018;73(suppl_2):ii2–ii10. 10.1093/jac/dkx504 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref008] 8.Department of Health. UK 5 Year Antimicrobial Resistance Strategy 2013 to 2018. London: Department of Health; 2013. [Google Scholar]

[pmed.1003202.ref009] 9.The Review on Antimicrobial Resistance by Jim O’Neill. Tackling drug-resistant infections globally: final report and recommendations. London: Review on antimicrobial resistance; 2016. [cited 2020 Jul 8]. Available from: https://amr-review.org/. [Google Scholar]

[pmed.1003202.ref010] 10.Gulliford MC, Moore MV, Little P, Hay AD, Fox R, Prevost AT, et al. Safety of reduced antibiotic prescribing for self limiting respiratory tract infections in primary care: cohort study using electronic health records. BMJ. 2016;354:i3410 10.1136/bmj.i3410 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref011] 11.Gulliford MC, Juszczyk D, Prevost AT, Soames J, McDermott L, Sultana K, et al. Electronically delivered interventions to reduce antibiotic prescribing for respiratory infections in primary care: cluster RCT using electronic health records and cohort study. Health Technol Assess. 2019;23(11):1–70. 10.3310/hta23110 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref012] 12.Cabral C, Lucas PJ, Ingram J, Hay AD, Horwood J. “It's safer to …” parent consulting and clinician antibiotic prescribing decisions for children with respiratory tract infections: An analysis across four qualitative studies. Soc Sci Med. 2015. [cited 2020 Jul 8];136–137:156–64. Available from: 10.1016/j.socscimed.2015.05.027. [DOI] [PubMed] [Google Scholar]

[pmed.1003202.ref013] 13.Lucas PJ, Cabral C, Hay AD, Horwood J. A systematic review of parent and clinician views and perceptions that influence prescribing decisions in relation to acute childhood infections in primary care. Scand J Prim Health Care. 2015;33:11–20. 10.3109/02813432.2015.1001942 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref014] 14.van den Broek d'Obrenan J, Verheij TJM, Numans ME, van der Velden AW. Antibiotic use in Dutch primary care: relation between diagnosis, consultation and treatment. J Antimicrobial Chemotherapy. 2014;69:1701–7. 10.1093/jac/dku005 [DOI] [PubMed] [Google Scholar]

[pmed.1003202.ref015] 15.Lusini G, Lapi F, Sara B, Vannacci A, Mugelli A, Kragstrup J, et al. Antibiotic prescribing in paediatric populations: a comparison between Viareggio, Italy and Funen, Denmark. Eur J Public Health. 2009;19:434–8. 10.1093/eurpub/ckp040 [DOI] [PubMed] [Google Scholar]

[pmed.1003202.ref016] 16.Laxminarayan R, Matsoso P, Pant S, Brower C, Røttingen J-A, Klugman K, et al. Access to effective antimicrobials: a worldwide challenge. The Lancet. 2016. [cited 2020 Jul 8];387:168–75. Available from: 10.1016/S0140-6736(15)00474-2. [DOI] [PubMed] [Google Scholar]

[pmed.1003202.ref017] 17.NHS England. Quality Premium: 2016/17 Guidance for CCGs. Leeds: NHS England; 2016 [cited 2020 Jun 1]. Available from: https://www.england.nhs.uk/wp-content/uploads/2016/03/qualty-prem-guid-2016-17.pdf.

[pmed.1003202.ref018] 18.NHS Choices. Sepsis. London: NHS; 2020 [cited 2020 Jul 8]. Available from: http://www.nhs.uk/Conditions/Blood-poisoning/Pages/Introduction.aspx.

[pmed.1003202.ref019] 19.National Institute for Health and Care Excellence. Sepsis: recognition, diagnosis and early management. NICE guideline [NG51]. London: NICE; 2017. Available from: https://www.nice.org.uk/guidance/ng51. [PubMed]

[pmed.1003202.ref020] 20.Royal College of Physicians. National Early Warning Score (NEWS) 2. London: Royal College of Physicians; 2017 [cited 2020 Jul 8]. Available from: https://www.rcplondon.ac.uk/projects/outputs/national-early-warning-score-news-2.

[pmed.1003202.ref021] 21.The Lancet Respiratory Medicine. Crying wolf: the growing fatigue around sepsis alerts. Lancet Respiratory Medicine. 2018;6:161 10.1016/S2213-2600(18)30072-9 [DOI] [PubMed] [Google Scholar]

[pmed.1003202.ref022] 22.Herrett E, Gallagher AM, Bhaskaran K, Forbes H, Mathur R, van Staa T, et al. Data Resource Profile: Clinical Practice Research Datalink (CPRD). Int J Epidemiol. 2015;44:827–36. 10.1093/ije/dyv098 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref023] 23.Herrett E, Thomas SL, Schoonen WM, Smeeth L, Hall AJ. Validation and validity of diagnoses in the General Practice Research Database: a systematic review. Br J Clin Pharmacol. 2010;69:4–14. 10.1111/j.1365-2125.2009.03537.x [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref024] 24.Gulliford MC, Sun X, Charlton J, Winter JR, Bunce C, Boiko O et al. Serious bacterial infections and antibiotic prescribing in primary care. Cohort study using electronic health records in the UK. BMJ open. 2020;10:e036975 10.1136/bmjopen-2020-036975 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref025] 25.Sun X, Gulliford M. Reducing antibiotic prescribing in primary care in England from 2014 to 2017: Population-based cohort study. BMJOpen. 2019;9:e023989 10.1136/bmjopen-2018-023989 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref026] 26.Clegg A, Bates C, Young J, Ryan R, Nichols L, Ann Teale E, et al. Development and validation of an electronic frailty index using routine primary care electronic health record data. Age and Ageing. 2016;45:353–60. 10.1093/ageing/afw039 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref027] 27.Gafoor R, Charlton J, Ravindrarajah R, Gulliford MC. Importance of Frailty for Association of Antipsychotic Drug Use with Risk of Fracture: Cohort Study Using Electronic Health Records. J Am Med Dir Assoc. 2019;20:1495–1501.e1. 10.1016/j.jamda.2019.05.009 [DOI] [PubMed] [Google Scholar]

[pmed.1003202.ref028] 28.Winter J, Charlton J, Ashworth M, Bunce C, Gulliford MC. Peritonsillar abscess and antibiotic prescribing for respiratory infection in primary care. Population-based cohort study and decision analytic model. Ann Fam Med. Forthcoming 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref029] 29.Gelman A, Carlin JB, Stern HS, Rubin DB. Bayesian Data Analysis. 3rd ed. Boca Raton, FL: Chapman and Hall / CRC; 2013. [Google Scholar]

[pmed.1003202.ref030] 30.Rhee C, Murphy MV, Li L, Platt R, Klompas M. Comparison of trends in sepsis incidence and coding using administrative claims versus objective clinical data. Clin Inf Dis 2015;60:88–95. 10.1093/cid/ciu750 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003202.ref031] 31.Balinskaite V, Bou-Antoun S, Johnson AP, Holmes A, Aylin P. An Assessment of Potential Unintended Consequences Following a National Antimicrobial Stewardship Program in England: An Interrupted Time Series Analysis. Clinical Infectious Diseases. 2019;69:233–42. 10.1093/cid/ciy904 [DOI] [PubMed] [Google Scholar]

[pmed.1003202.ref032] 32.Gharbi M, Drysdale JH, Lishman H, Goudie R, Molokhia M, Johnson AP, et al. Antibiotic management of urinary tract infection in elderly patients in primary care and its association with bloodstream infections and all-cause mortality: population-based cohort study. BMJ. 2019;364:l525 10.1136/bmj.l525 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Probability of sepsis after infection consultations in primary care in the United Kingdom in 2002–2017: Population-based cohort study and decision analytic model

Martin C Gulliford

Judith Charlton

Joanne R Winter

Xiaohui Sun

Emma Rezel-Potts

Catey Bunce

Robin Fox

Paul Little

Alastair D Hay

Michael V Moore

Mark Ashworth

Roles

Abstract

Background

Methods and findings

Conclusions

Author summary

Why was this study done?

What did the researchers do and find?

What do these findings mean?

Introduction

Methods

Ethics statement

Data source

Sepsis events

Selection of sample for antibiotic prescribing analysis

Evaluation of frailty

Decision tree

Fig 1. Decision tree showing the probability of a patient consulting for an infection, being prescribed an antibiotic at that consultation, and developing sepsis.

Table 1. Definition and data source for probabilities.

Sensitivity analyses

Results

Table 2. First sepsis events recorded in CPRD from 2002 to 2017 and preceding infection consultations and AB prescriptions.

Table 3. Probability of sepsis after infection consultations in primary care.

Fig 2. Probability of sepsis following infection consultations in primary care if ABs are prescribed or not (upper panel).

Fig 3. Number of antibiotic prescriptions required to prevent one sepsis event (NNT) following infection consultations in primary care by frailty level, gender, and age group.

Fig 4. Number of antibiotic prescriptions required to prevent one sepsis event (NNT) by age group, gender, and type of infection consultation.

Sensitivity analyses

Discussion

Main findings

Strengths and limitations

Comparison with other studies

Conclusions

Supporting information

Acknowledgments

Abbreviations

Data Availability

Funding Statement

References

Decision Letter 0

Artur Arikainen

Roles

Decision Letter 1

Artur Arikainen

Roles

Author response to Decision Letter 1

Decision Letter 2

Artur Arikainen

Roles

Author response to Decision Letter 2

Decision Letter 3

Artur Arikainen

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases