Perioperative intravenous fluid prescribing: a multi-centre audit

This was a large retrospective audit of current intravenous fluid prescribing practice
in the five hospitals in the Wessex region. We found that early cessation of intravenous
fluids in the postoperative elective surgical patient was common and in keeping with
the approach promoted by enhanced recovery after surgery programmes. In those patients
that continue to receive intravenous fluids, a considerable proportion appear to receive
excessive or inadequate quantities of various electrolytes in relation to reference
requirements. A small but important proportion appears to receive excessive volumes
of water. The type of intravenous fluid administered varies according to the stage
of the patient pathway. There was considerable variation in the volume of fluid administered
in the operating theatre for three common elective procedures. Electrolyte disturbances
were frequent and became more common as the postoperative days passed.

Strengths of this audit

This is a large, multi-centre audit across five hospitals serving a wide geographical
area with a sample population typical of most regions across the country. The included
surgical procedures are common and relevant to a wide audience. The data collected
has allowed a detailed analysis of the current practice, and the findings are in keeping
with previous work (Minto and Mythen 2015]; Callum et al. 1999]; Lilot et al. 2015]). The majority of recent major trials regarding intravenous fluids focus on intraoperative
fluid management. However, only a small proportion of the perioperative journey is
spent in the operating theatre under the direct care of an anaesthetist. Our work
addresses both the intraoperative and postoperative periods.

Limitations of this audit

Heterogeneity

The patients included are a very heterogeneous group, with a wide range of age, body
weight, mode of anaesthesia and operation type.

Data capture

For administrative and logistical reasons at the different institutions, we could
not obtain 100 % of the notes identified leading to a potential for selection bias.
Hospital C contributed relatively more patients potentially skewing the overall results.

Data limitations

The majority of medical records in this audit were paper based and often subjectively
of poor quality which may affect the results. We did not collect detailed data on
individual patient perioperative risk scores in the form of ASA (American Society
of Anaesthesiologists) grade or other perioperative risk scoring systems. We considered
ASA grade to be insufficiently robust to determine perioperative risk, and other risk
scoring systems were rarely used by anaesthetists in this audit.

When calculating intraoperative fluid administration data for the three common procedures,
our ability to determine that these were uncomplicated was based on discharge summaries
and anaesthetic charts. Neither of these two methods may have recorded all the technical
difficulties or other ‘complications’ during the surgery that may have mandated higher
than average intravenous fluid administration volumes. In addition, the length of
time spent in the operating theatre was not recorded so we were unable to present
intraoperative fluid administration volumes as ml.kg.h
^?1
. Also, these patients were not matched in terms of perioperative risk score (e.g.
ASA), co-morbidities, surgical complexity or operative time, although we did exclude
non-elective patients and those with significant blood loss. Fluid balance charts
were studied to help assess appropriateness of intravenous fluids. These were subjectively
noted to be often of poor quality and do not record all losses, such as fluid sequestration
in sepsis, so they do not provide all the information needed to assess whether the
fluid administered is clinically appropriate. Fluid prescription charts were better
in terms of completion.

Study design

Volumes of fluid and quantities of sodium were calculated to be excessive in reference
to the maintenance requirements in a large number of patients in the postoperative
period. An unknown proportion might have been septic, bleeding or hypovolaemic requiring
additional replacement over and above the maintenance requirements. A detailed notes
review would be required to assess how clinically appropriate the fluid volume and
sodium quantity given in each case was. Enhanced recovery pathways and fluid prescription
guidelines often vary between individual hospitals and even individual surgical teams
in the same institution. This may explain some of the variation in practice.

Meaning of this audit

One of the principles of enhanced recovery after surgery protocols is an early return
to oral intake to improve patient comfort, gut function and limit the detrimental
effects of intravenous fluid (Guidelines for the Implementation of Enhanced Recovery
Protocols 2009]). Intravenous fluid guidelines (National Institute for Health and Care Excellence
2013]; Powell-Tuck et al. 2011]) also make reference to promoting the early return to oral intake. Over 50 % of the
patients had a free oral fluid order on the day of their operation or postoperative
day 1, indicating an attempt to return them to normal oral fluid intake. Of the patients,
32 % had no record of any free fluid order being made possibly because it was obvious
that the patient could drink freely or that the hospitals have adopted an enhanced
recovery approach and assumed free oral fluids can be given unless specifically documented
otherwise. Poor medical record keeping is another possibility. At least some of these
32 % of patients would have been allowed to drink free oral fluids; therefore, the
lack of documentation is unlikely to underestimate the already large number of patients
with an early (day of operation or postoperative day 1) free fluid order.

The balanced crystalloid solutions were the most common postoperative fluid, with
hypotonic crystalloids (dextrose-saline solutions) and 0.9 % sodium chloride as the
next most common. There is increasing recognition that the traditional postoperative
regime of 0.9 % sodium chloride and 5 % dextrose risks sodium, chloride and salt overload
(De Silva et al. 2010]). Sodium chloride (0.9 %), even when not given in excessive quantities, is associated
with a variety of detrimental effects such as hyperchloraemic acidosis, reduced renal
blood flow, increased chance of renal failure and increased in-hospital mortality
after major abdominal surgery (Lobo 2012]). Data from Wessex in 2007 found that over 70 % of postoperative fluid prescriptions
were 0.9 % sodium chloride or 5 % dextrose (De Silva et al. 2010]). This figure was reduced to 40 % in 2009 after a targeted education intervention
(De Silva et al. 2010]). It appears that this downward trend has continued as evidenced by our work.

Differences in intraoperative and postoperative fluid types is partially explained
by the prescriber (anaesthetists in theatre vs junior doctors or non-anaesthetists
postoperatively) and the different physiology occurring at the different time points.

Even the use of the balanced crystalloids can result in sodium overload and some hospitals
and the NICE (National Institute for Health and Care Excellence 2013]) advocate hypotonic dextrose-saline solutions as a means of giving water with minimal
sodium to meet the maintenance requirements. It is not surprising that the greater
use of hypotonic dextrose-saline solutions was seen in patients that went to the ICU
postoperatively compared to HDU or ward, perhaps indicating better intravenous fluid
prescribing practice or awareness of guidelines.

The inadequate doses of intravenous potassium required to meet the maintenance seen
in our audit are in keeping with previous work (Lu et al. 2013]) and our anecdotal observations. Although it is possible that some patients received
potassium supplementation by other means (i.e. orally), the increasing incidence of
hypokalaemia as the postoperative period progresses suggests that inadequate intravenous
(and other) potassium supplementation is a true finding and is a problem that requires
addressing. There are a number of potential reasons behind this but fear of intravenous
potassium and the belief that the balanced crystalloid solutions contain adequate
potassium for all patients are possible explanations. Hypokalaemia in the postoperative
patient has been associated with slower return of gut function as well as other complications,
and there is a suggestion that preventing hypokalaemia in the postoperative stage
may improve outcomes (Lu et al. 2013]).

Postoperative hyponatraemia was common and is multifactorial. It is not possible to
determine the causes without a detailed notes analysis of the affected patients. As
Fig. 8 demonstrates, mean serum sodium declined by 2.6 mmol.l
^?1
in the 356 subjects in whom pre- and postoperative day 1 serum sodium values were
available. This is despite the vast majority of intraoperative intravenous fluid therapy
consisting of balanced crystalloid solutions. Therefore, the decline in mean serum
sodium compared to pre-operative values is most likely to represent the physiological
stress response to surgery. This is a spectrum of changes that occur throughout various
body systems (neuroendocrine, metabolic, immunological and haematological) in response
to surgical incision and trauma. Neuroendocrine changes are particularly relevant
to perioperative fluids because the release of catecholamines and cortisol, vasopressin
and aldosterone result in retention of sodium and water (often water in excess of
sodium hence hyponatraemia), loss of potassium, reduced creatinine clearance and urine
output. These effects can last well beyond the operative period into the postoperative
phase.

Hyponatraemia can cause a variety of neurological and other symptoms but, more importantly,
has been associated with increased risk of in-hospital and long-term mortality in
a variety of patient groups (Lu et al. 2013]). The mean drop in serum sodium from postoperative day 1 to postoperative day 2 was
0.21 mmol.l
^?1
. When only looking at those patients who received any dextrose solutions on postoperative
day 1, the mean drop was 2.9 mmol.kg
^?1
. This raises the possibility that the use of dextrose solutions on postoperative
day 1 is associated with subsequent hyponatraemia (compared to other fluids) and should
be avoided. It should be noted that the pre-operative mean serum sodium was lower
in this group. With the relatively small number of subjects, the retrospective nature
of the audit and the large number of contributory factors, it is not possible to suggest
causation. From our limited data, it would appear that the use of hypotonic dextrose-saline
on postoperative day 1 was not associated with hyponatraemia and therefore, the recommendation
made in several fluid guidelines (National Institute for Health and Care Excellence
2013]; Woodcock 2014]) that this is the preferred fluid is reasonable based on the results of this audit.

The range of intraoperative fluid doses for three index procedures was wide and in
keeping with a recent two-centre observational study in America of fluid administration
for uncomplicated elective abdominal surgery with minimal blood loss (Lilot et al.
2015]). This found wide variability of crystalloid administration both within and between
anaesthesia providers (Lilot et al. 2015]). In that study, for most procedures, 50 % of patients received 4–10 ml.kg
^?1
.h
^?1
of crystalloid (corrected for urine output and estimated blood loss) but 50 % fell
outside of this wide range with some patients receiving as much as 35 ml.kg
^?1
.h
^?1
(Lilot et al. 2015]). Although a patient’s fluid requirements will vary and depend on a number of factors,
it would be surprising to see such a wide range of physiological needs during similar
surgical episodes, raising the possibility that variation is due to variation in individual
anaesthetists’ fluid approach.

The lack of correlation between the volume of postoperative day 1 fluids and body
weight is interesting. In terms of absolute volumes given, if clinicians were calculating
fluid doses on a millilitre per kilogram basis, we might expect absolute volumes to
increase with body weight. We saw the opposite of this with smaller patients getting
larger volumes of fluids. This suggests that practitioners are thinking in terms of
‘number of bags’ for a ‘standard’ patient and not taking into account weight.

Unanswered questions and future research

Most intravenous fluid trials take place in the operating theatre. We have demonstrated
the type of intravenous fluid prescribed varies depending on whether the patient is
in the postoperative ward or in theatre. The type of prescriber is likely to be different
as well. Some previous intravenous fluid research has excluded the first postoperative
day from any analysis because fluid prescriptions during this period may have been
complicated by surgery-induced fluid and electrolyte shifts and any postoperative
fluid prescriptions completed by the anaesthetist (Walsh and Walsh 2005]). Being a continuum, any future research should cover the intraoperative and postoperative
phases, something that has been recognised as important by an international trial
looking at fluid therapy during and after major abdominal surgery Myles PS Wallace
SK (2015]). Interestingly, prostatectomies in the American study (Lilot et al. 2015]) had a much narrower range of fluid administered because a fluid protocol exists.
A similar fluid protocol intervention may be needed in Wessex.

Education of junior doctors can make an impact on their prescribing choices (De Silva
et al. 2010]) but there is always a risk that any relatively short lived intervention only lasts
a limited period of time before traditional practice starts to re-establish itself.
The variation in intraoperative fluid volumes observed in theatre means that there
is a quality improvement work that might be required in this environment as well.