Waterlow score for risk assessment in surgical patients: a systematic review

Abstract

Introduction

The Waterlow score (WS) is used routinely in clinical practice to assess risk of pressure sore development. Recent studies have also suggested its use in preoperative risk stratification. The primary aim of this systematic review was to evaluate the current evidence on the WS in predicting morbidity and mortality in surgical patients.

Methods

A systematic review was carried out in accordance with PRISMA and SWiM guidelines. A search strategy was conducted on the MEDLINE and EMBASE databases. Quality was assessed using the Newcastle–Ottawa scale.

Findings

Overall, 72 papers were identified, of which 7 met inclusion criteria for full text review, and 4 were included for analysis. All studies were cohort in nature and published between 2013 and 2016, encompassing a total of 505 surgical patients. The studies included general, vascular, transplant and orthopaedic surgery. A high WS was demonstrated to have statistically significant association with increased morbidity and mortality as well as need for intensive care unit admission and length of stay. Furthermore, this was a more accurate predictor compared with the P-POSSUM and ASA scoring systems used currently in routine practice.

Conclusions

The WS is a promising tool for risk stratification of surgical patients. It is already collected routinely by nursing staff throughout hospitals in the UK and would therefore be easy to implement. However, further large prospective studies are required in order to validate these findings prior to its establishment for this role in everyday surgical practice.

Introduction

As the global volume of surgery continues to rise, there is a concomitant increase in the impact from perioperative morbidity and mortality.^1,2 In order to optimise outcomes from surgery it is essential that we are able to characterise those who are most at risk of complications. This principle also applies to patients with surgical diagnoses, such as pancreatitis or cholecystitis, who may be managed with conservative measures. Clinical judgement alone has been demonstrated to be an unreliable predictor of adverse outcomes.^3,4 Therefore, the use of validated risk stratification tools can aid in identification of patients who are at increased risk of perioperative morbidity or mortality and may benefit from targeted interventions.^3,4

Routinely collected hospital data are not standardised.⁵ A number of generic and specific assessment tools exist in order to predict the risk of postoperative complications. Most notably, the American Society of Anaesthesiologists (ASA) physiological status classification is used as a convenient tool. This scale, however, has been demonstrated to have poor interobserver reliability, limiting its specificity and sensitivity.^6–9

The Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity (P-POSSUM) and Acute Physiology and Chronic Health Evaluation II (APACHE II) scores are examples of detailed tools in surgery and intensive care. One disadvantage of the APACHE II scoring system is that it requires measurement of several variables, rendering it somewhat complex.¹⁰ Indeed, assessment tools that are scenario-specific lack clinical versatility. An ideal risk predictor should rely on routinely captured clinical data, be easily accessible, be validated and applicable to a range of scenarios. Its performance characteristics should include high sensitivity and specificity. To date, there is no risk prediction system that satisfies all of the above criteria.¹¹

The Waterlow score (WS) is a validated composite tool that is commonly used by nursing staff in order to stratify risk of pressure sore development.^12,13 Efficacy of this tool as a predictor of morbidity and mortality of surgical patients has been studied, and its utility in predicting additional clinical indices such as length of hospital stay (LOS) have been demonstrated.¹⁴ This tool is weighted for patient gender, age group, body mass index (BMI), skin health, mobility, continence, nutrition and ‘special risk’ factors.^15,16 A score of 20 or more indicates very high risk of ulcer development, whereas a score of 9 or less represents little or no risk.¹⁶ A key advantage of this tool is that it is used routinely throughout UK hospitals and patients are re-evaluated on a weekly basis, providing a dynamic risk stratification score.

The primary objective of this systematic review was to assess the current evidence on the WS as a universal risk assessment tool for morbidity and mortality in surgical patients regardless of the modality of treatment, whether operative or conservative.

Methods

Protocol and registration

A systematic review of the published literature was carried out in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) and Synthesis without meta-analysis (SWiM) guidelines.^17,18 The review protocol was registered in PROSPERO, the international prospective register of systematic reviews database (registration number CRD42020188252).

Eligibility criteria

Inclusion criteria were studies assessing the role of the WS as a risk assessment tool for surgical morbidity and mortality. Exclusion criteria were conference abstracts, opinion-based reports and publications relating to pressure sores.

Information sources and search strategy

A comprehensive search of the published literature on the MEDLINE and EMBASE databases from inception to 24 May 2020 was carried out. The search was triangulated with a search of the bibliography of the included studies as well as recent reviews on the WS. The following search terms were used: (‘Waterlow’) AND (‘surg*’ OR ‘operat*’).

Study selection

Two reviewers independently performed eligibility assessment. This was initially carried out through screening of article titles and abstracts; the process was completed by full text evaluation. Disagreements between reviewers were resolved by consensus with the senior author.

Data collection process and data items

Information was extracted onto a standardised proforma. Information collected from each paper consisted of level of evidence, study design, patient criteria, sample size, timing of WS collection, WS values, postoperative complications and statistical tests used.

Summary measures and synthesis of results

Data analysis was carried out quantitatively looking at odds ratios (OR) with 95% confidence intervals (CI). Result heterogeneity was evaluated to see if meta-analysis was possible. Furthermore, qualitative data such as author discussion and conclusions were considered for narrative synthesis of findings.

Assessment of data quality

Risk of bias was assessed for each individual study. This was carried out by two researchers using the Newcastle–Ottawa scale.¹⁹ Each study was given a score out of 9 based on three broad categories: selection, comparability and outcome.

Results

Study selection

The initial search identified 72 articles, of which 25 were duplicates. When screening the titles and abstracts, 40 failed to meet the inclusion criteria. The remaining seven articles were retrieved for full text review. From these, three were excluded; the remaining four papers were included for analysis.^20–23 A flow diagram of study selection is illustrated in Figure 1. Due to the heterogeneity in outcome measures between studies, a meta-analysis was not feasible.

Figure 1 .

Flow diagram of study selection

Study characteristics

A list of the included studies and a summary of their characteristics are presented in Table 1. The eligible studies were from four units in the UK and published between 2013 and 2016, encompassing a total of 505 surgical patients. All four were cohort studies, of which one was prospective and the remaining retrospective.²² Two studies reported on general surgery,^20,23 one on orthopaedic surgery²¹ and one on transplant surgery.²² Different cut-off points of WS, ranging from 15 to 20, were reported in the eligible studies, permitting categorisation of patients into high risk and low risk.

Table 1 .

Study characteristics

Study ID	Author	Published year	Study period	Study design	Quality Assessment	Surgery type	Study location	Sample size	Mean age (year)	Timing of WS	WS cut-off point for ‘high-risk'	Outcomes	OR	95% CI (LL-UL)
1	Thorn et al20	2013	2010	RC	8	General or vascular	Stevenage, UK	101	68 (median)	Preoperative	20	Inpatient mortality	14	2.5	8.3
												30-day morbidity	15	1.7	38
2	El-Daly et al21	2015	2010–2011	RC	9	Neck of femur fracture	London, UK	97	75	Preoperative	17	30-day infection	36.17	9.61	136.15
3	Khambalia et al22	2015	2011–2013	PC	8	Simultaneous pancreas and kidney transplant (recipients)	Manchester, UK	57	42	Preoperative	N/A	Total LOS	N/A	N/A	N/A
												Critical Care LOS	N/A	N/A	N/A
4	Gillick et al23	2016	2010–2013	RC	7	Acute pancreatitis	Exeter, UK	250	66.6 (median)	Preoperative	15	Inpatient mortality	7.23	2.73	19.14
												ICU admission	3.68	1.41	9.58
												LOS >7 days	3.11	1.7	5.71

CI = confidence interval; LL = lower limit; LOS = length of stay; N/A = not applicable; OR = odds ratio; PC = prospective cohort; RC = retrospective cohort; UL = upper limit; WS = waterlow score

Quality assessment was done using the Newcastle–Ottawa Scale (range 0–9) for non-randomised observational studies

Outcomes

Mortality

Two studies in general surgery patients, one looking at patients with acute pancreatitis (n=250) and the other a variety of patients who underwent emergency or elective general surgical operations (n=101), found that a higher WS was associated with significantly higher inpatient death (OR: 7.23, 95% CI: 2.73–19.14 and OR: 15, 95% CI: 1.7–38, respectively).^20,23 Mortality was not assessed as an outcome measure in the remaining two studies due to the low mortality rates in their patient cohorts.

LOS in hospital/intensive care

One study in simultaneous pancreas and kidney transplant recipients (n=57) demonstrated WS to be a significant predictor for total LOS and intensive care unit (ICU) LOS (p<0.001).²² Similarly, another study indicated that in patients with acute pancreatitis, those with a high WS (>15) have a higher risk (OR: 3.11, 95% CI: 1.70–5.71) of a LOS longer than seven days (p=0.0002).²³ In addition, studies suggested that a higher preoperative WS is associated with a higher chance of ICU admission and 30-day postoperative morbidity.^20,23

Postoperative infections

One study among patients who had a neck of femur fracture (n=97) found that a high WS (>17) is associated with higher 30-day postoperative infection rate (OR: 36.17, 95% CI: 9.61–136.15) and with every one point increase in WS, there is a 1.68-fold increase in the odds of developing postoperative infection.²¹

Comparison of WS with other validated tools

Thorn et al compared the receiver operating characteristics (ROC) for mortality, with similar predictive values demonstrated between the WS (0.81), P-POSSUM (0.85) and ASA (0.80).²⁰ A similar finding was noted for morbidity.

Khambalia et al evaluated the P-POSSUM, multiple organ dysfunction score, Charlson comorbidity index and revised cardiac risk index as predictors of postoperative outcome in simultaneous pancreas and kidney transplantation (n=57), but a correlation between the tools was not performed.²² These scores were calculated on admission prior to surgery from which WS was the only significant predictor of outcomes, namely total LOS and ICU LOS.²²

In the study of acute pancreatitis, WS significantly correlated with Glasgow score (p=0.0012).²³ They also demonstrated WS to be comparable with recognised markers of pancreatitis severity, namely CRP at 48h and Glasgow score, with regard to ROC for mortality (area under the curve (AUC) 0.73 vs 0.66 vs 0.62, respectively), ICU admission (AUC 0.65 vs 0.87 vs 0.68, respectively) and LOS >7 days (AUC 0.64 vs 0.79 vs 0.65, respectively).²³

Risk of bias

All of the included studies demonstrated a high quality of methodology, scoring 7 to 9 on the Newcastle–Ottawa scale (Table 2).

Table 2 .

Newcastle–Ottawa scale for the quality assessment of included cohort studies

Author	Year	SELECTION				COMPARABILITY		OUTCOME		Total score
		Representativeness of the exposed cohort	Selection of the non exposed cohort	Ascertainment of exposure	Outcome was not present at start of study	Control for important or additional factors	Assessment of outcome	Long follow up for outcomes to occur	Adequacy of follow up of cohorts
Thorn et al20	2013	⋆	⋆	⋆	⋆		⋆	⋆	⋆	7
Khambalia et al21	2015	⋆	⋆	⋆	⋆	⋆	⋆	⋆	⋆	8
El-Daly et al22	2015	⋆	⋆	⋆	⋆		⋆	⋆	⋆	7
Gillick et al23	2016	⋆	⋆	⋆	⋆	⋆	⋆	⋆	⋆	8

Discussion

The primary aim of this systematic review was to assess the current evidence on the WS as a risk assessment tool for morbidity and mortality in surgical patients. The WS is used almost universally across NHS trusts for risk stratification of pressure sore development. Compliance rates close to 100% are associated with use of WS.^14,20 Use of an existing infrastructure for risk prediction in surgery avoids the costs and barriers of implementing a new tool. The WS was designed to enhance education and optimise resource management in decubitus ulcer management.¹² This review suggests that use of the WS as a predictive tool in surgery would be a resource neutral initiative, given its near universal use in secondary care.

The ability of WS to predict postoperative complications was first reported by Thorn et al, who demonstrated a statistically significant association between preoperative WS and inpatient mortality and 30-day morbidity in elective and emergency general and vascular surgery. Furthermore, they found the WS to have similar predictive value to the P-POSSUM and ASA scoring systems.²⁰

Khambalia et al demonstrated that, in the setting of pancreas and kidney transplantation, high preoperative WS was a strong predictor of total and ICU LOS, which can be considered a surrogate maker of postoperative morbidity.^24,25 Indeed other risk prediction tools, such as the P-POSSUM and ASA, did not yield similar results.²² To date, no other accurate risk stratification scoring systems have been validated for use in transplant surgery.²²

In patients with hip fractures, El-Daly et al found the WS to be an accurate predictor of postoperative infection within 30 days, regardless of procedure performed. Increasing time to surgery, age and comorbidities such as diabetes mellitus have all been previously linked to increased risk of postoperative infection for neck of femur fractures; these latter two parameters are included in the WS.^26–28 Hence, the WS may be a useful marker for identifying high-risk patients who might require an alternative antibiotic prophylaxis regime.²¹ Furthermore, given that the WS has been designed to identify patients vulnerable to skin breakdown, this may be applicable across surgical specialities and warrants further research.

Gillick et al demonstrated that, in patients with acute pancreatitis, WS on admission was an accurate predictor of mortality, ICU admission and LOS. This was comparable with the predictive power of the Glasgow score on admission for assessment of pancreatitis disease severity.²³ It should be noted, however, that the Glasgow score is a well-established and validated tool for pancreatitis severity, while these findings are from a single-centre retrospective study. Further research is therefore required to validate the results from this study.

Based on the above evidence, the WS can be considered as a potential candidate for risk stratification in surgery. An interesting aspect of this review is that WS appears to predict outcomes in a range of clinical settings from conservative management of acute pancreatitis, hip fracture to pancreatic transplantation. Of note, the WS encompasses a number of parameters that link to frailty, and can therefore be considered as a surrogate frailty index, which has also been demonstrated as a predictor of mortality in emergency general surgery admissions.^29,30 Encouragingly, the largest study to evaluate WS as a predictor of 30-day mortality in acutely unwell ‘medical’ patients confirmed that WS is a significant predictor of this metric with OR 1.12 and AUC of 0.69.¹⁴

Study limitations

There are a number of limitations to be considered. To date, the limited evidence regarding WS in surgical patients is harnessed largely from retrospective cohort studies. It is conceivable that studies with ‘negative’ findings have not been published. Therefore, large prospective studies are required in order to validate these findings prior to adopting this as a surgical risk predictor.

Furthermore, studies have demonstrated the WS to have variable inter-rater reliability due to the presence of subjective variables.^31,32 Therefore, further comparative studies are required against existing more objective tools such as the P-POSSUM. This would be an eminently suitable subject for evaluation at scale by a national trainee research collaborative.³³ Such a collaborative can be carried out prospectively across surgical specialties looking at outcomes including length of stay, surgical morbidity and mortality. While the WS has the advantage of near universal use in nursing practice, it may not be suitable for immediate or urgent emergency cases due to the lag between admission and pressure ulcer assessment. These cases may therefore need to be excluded from any collaborative project.

Conclusions

This systematic review has demonstrated that the WS could be an effective predictor for risk of morbidity and mortality in selected surgical patients. Further research with large prospective studies is required in order to validate these findings. Following this, there may be a role for the WS to be established as a risk stratification tool for surgical patients on admission to hospital, which would be easy to implement given its familiarity among hospital staff.