Acute paediatric units require round-the-clock skilled resident medical cover. Fully trained doctors remaining resident on-site at night and weekends may improve care at these times, but costs are higher. In compensation, more senior doctors may be less likely to admit children.
Methods In a unit providing 24-hour, 7-day acute services, out-of-hours resident cover has been divided between level 2/3 trainees and consultants. Between 2007 and 2017, night and weekend day shifts were identified as resident consultant or non-resident consultant. Admission numbers (duration of stay of ≥4 hours) were obtained from hospital activity databases. Analyses were undertaken on total admissions and stratified by time of day and duration of stay of >12 or < 12 hours. Incidence rate ratios (IRRs) were derived using negative binomial regression .
Results For all out-of-hours and short-stay patients, children were significantly more likely to be admitted when there was no resident consultant: IRRs 1.07 (95% CI 1.04 to 1.09) and 1.09 (95% CI 1.02 to 1.18), respectively. There was no difference between rates stratified into long stay at night or weekend days: IRRs 1.01 (95% CI 0.96 to 1.07) and 1.03 (95% CI 0.99 to 1.18) respectively .
Conclusion A resident consultant presence was associated with reduced total, night-time and short-stay admissions.
- Paediatric Practice
- Paediatric Staffing
Statistics from Altmetric.com
What is already known on this topic?
Limited availability of skilled trainee paediatricians creates difficulties in staffing acute rotas out-of-hours.
Fully trained paediatricians remaining resident in hospitals out-of-hours may offer a solution, but employment costs are higher compared with trainees.
What this study adds?
Numbers of acute paediatric admissions are marginally reduced during out-of-hours shifts where the most senior resident doctor is a consultant, compared with shifts where it is a trainee.
The effect is most obvious at night, and in less ill children.
Acute paediatric units within the UK National Health Service require round-the-clock skilled resident medical cover, but the numbers of level 2 and 3 trainee doctors able to provide this are insufficient, given working-time restrictions. Increasingly, fully-trained doctors are being used to provide this by remaining resident on-site at night and during weekend days. This policy has been promoted by a report from the Academy of Medical Royal Colleges and through standards published by the Royal College of Paediatrics and Child Health.1 2 Although early consultant review is likely to enhance the quality of care received by children being assessed at these times, the costs to providers are higher than when trainees provide the cover. If consultants were to admit fewer children, there might be cost savings. We sought to compare admission rates, depending on whether or not a consultant was resident.
Salisbury District Hospital is a small district general hospital in the south of England. It provides a 24-hour, 7-day acute paediatric and neonatal service to an under-16 population of around 40 000. The service includes a 16-bed children’s ward, a level 2 neonatal unit and a maternity unit with around 2300 deliveries per annum. From 2007, the out-of-hours resident cover was split between four level 2/3 trainees (paediatric specialty trainees years 4–8), and a team of 6 consultants, later increasing to 10. Consultants stayed on site for the entire duration of the shift. At all times, there was a relatively inexperienced level 1 doctor also resident (foundation year 2, or specialty trainee years 1–3). Level 1 doctors were permitted to decide on admission without necessarily consulting a senior. The allocation of resident night and weekend day shifts was according to a fixed rolling rota and took no account of levels of activity or season. Distribution of resident consultant and trainee-only resident shifts was therefore effectively random.
Duty rotas between February 2007 and June 2017 were reviewed, and each night (Monday–Sunday 21:00 to 09:00) and weekend day (Saturday and Sunday 09:00–21:00) shift was identified as resident consultant or non-resident consultant. Hospital activity databases were interrogated to establish numbers of admissions under medical paediatrics. An admission was defined as a stay in hospital of 4 hours or more. For night shifts, admissions were included only from 21:30 onwards, 30 mins after the start of the shift, to allow for the probability that the decision in cases admitted before 21:30 would have been made by the day-shift doctor. However, this ‘handover’ exclusion was not made for weekend day shifts, as very few children were admitted between 08:00 and 09:00. In order to identify those who might have had a more significant illness, admissions were dichotomised into those that lasted more or less than 12 hours.
Data on diagnosis and treatment were not considered. It was not possible to exclude those who were transferred off-site for intensive care or for more specialist tertiary care. However, these were few in number (<5%) and would have been distributed evenly between consultant resident and non-consultant resident shifts, so should not affect the comparison. Patients admitted under a paediatric specialty but aged above 19 years or admitted to a non-paediatric ward were excluded, as these were likely to represent data quality errors.
The data were therefore grouped by resident consultant (present or not), time of day (weekend day or night) and length of stay (less than or more than 12 hours). This resulted in six comparable groups of data. Comparisons of rates of admissions per group were made using a negative binomial regression model. The models were run for total admissions and, in addition, stratified by time of day (09:00–21:00 or 21.30–09:00) and length of stay (long, >12 hours; short,<12 hours) for each variable. As the counts did not fulfil assumptions required for a Poisson model, negative binomial regression was used to generate incidence rate ratios (IRRs) and 95% confidence intervals (95% CIs).
All statistical analyses were undertaken using STATA software V.13.
Over the 11-year period, there were 2549 consultant resident nights and 1233 nights when no consultant was resident. Of the 6830 children admitted during night shifts, 4727 were admitted on resident consultant nights, and 2103 were admitted on non-consultant nights. There were 489 weekend days when consultants were residents and 586 when not. Of the 3205 children admitted on weekend days, 1403 were admitted when a consultant was resident, and 1802 when no consultant was resident.
There were 5957 (59.4%) admitted either at night or on weekend days that stayed more than 12 hours, suggesting that they had a more significant illness.
Comparisons of numbers of admissions on non-resident consultant shifts versus resident consultant shifts are listed in table 1 as IRRs. For all shifts, children were significantly more likely to be admitted on a non-resident consultant shift (IRR 1.07; 95% CI 1.04 to 1.09, p=0.0001).
When the shifts were stratified by nights and weekend days, and by short and long admission durations (table 1), the only significant effect occurred for short admissions at night: IRRs for all nights, long-stay nights, short-stay nights, all weekend days, long weekend days and long weekend nights, were respectively, 1.03 (95% CI 0.99 to 1.09), 1.01 (95% CI 0.96 to 1.07), 1.09 (95% CI 1.02 to 1.18), 1.03 (95% CI 0.99 to 1.18), 0.98 (95% CI 0.94 to 1.03) and 1.07 (95% CI 0.99 to 1.16).
We found a small but significant indication of greater likelihood of admission when a consultant was not resident, but this was statistically significant only for the whole group and the subgroup of shorter admissions during night shifts. This may be because less experienced doctors seeing children are more likely to be cautious and thus advise admission, while a consultant may feel confident in allowing the child to go home in similar circumstances. The observation that no significant differences were seen for those who stayed more than 12 hours and were thus likely to have been more unwell, suggests that the difference was largely explained by decisions made on less ill children.
In this study, handover-time admissions between 21.00 and 21.30 were excluded because it was impossible to ascertain whether decisions made during this period were by consultant or middle-grade doctors. This may have concealed a possible benefit of consultant resident shifts, that is, more appropriate decision-making following discussion with a consultant at evening handover. Some decisions by middle-grade doctors may have been influenced by telephone discussions with the non-resident consultant, but our data did not allow this to be assessed.
No formal comparison between the groups was made of age, sex and diagnosis, nor were any seasonal or secular trends assessed: as the distribution of resident consultant versus non-resident consultant shifts was effectively random throughout the entire period, there is no reason to suppose that these would have differed.
There has been little previous work that has set out to analyse objectively the benefits of senior doctors remaining resident on-site at nights and weekends. In paediatrics, one study which was not restricted to nights and weekends demonstrated marginal benefits from early consultant review on an acute unit in reducing length of stay, but only for specific subgroups of patients.3
The effect could be explored further by examining whether there is a gradient in admission rates depending on the seniority of the middle-grade doctor (specialty trainee year 4–8).
Future work might also examine numbers of investigations ordered, compare repeat presentation rates, and assess patient and parent satisfaction with receiving a prompt consultant assessment out-of-hours.
An economic analysis might indicate whether this reduction in admissions could justify the extra costs of employing consultants out-of-hours, but this was beyond the scope of this study.
In conclusion, this study showed some small benefits from resident consultants out-of-hours in terms of reducing admissions, although admission rates are not of themselves an indicator of quality of care.
The authors thank Dr Paul Strike for providing statistical advice.
Correction notice This article has been updated since it was published online. Please see the updated author note.
Contributors RS-J conceived the project, developed the research methods and wrote the text. EC collated and analysed the data, with statistical support. NB did further analysis for the resubmission and advised on the final text.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request. Data available from the authors.
Publisher note This article includes Editor-in-Chief Nick Brown as an author. To avoid a conflict of interest the Associate Editor Colin Powell handled the paper throughout the submission process.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.