## Abstract

Introduction: Inflammatory bowel disease (IBD) is a chronic condition, yet the model of care is often reactive. We sought to examine whether a formal IBD service (IBDS) reduced inpatient healthcare utilisation or lowered costs for inpatient care.

Material and methods: With protocols, routine nurse phone follow-up a help-line, more proactive care was delivered, with many symptoms and concerns dealt with prior to routine presentation. Over two five month periods before (2007/8) and after (2009/10) introducing a formal IBDS two discrete cohorts of admitted IBD patients were identified at a single centre. Each patient was assigned five contemporaneously admitted, age and gender matched controls. Inpatient healthcare utilisation was compared between patients and controls and disease-specific factors amongst the two IBD cohorts.

Results: The initial audit captured 102 admitted IBD patients (510 controls, median age 44 years, 57% female); the second audit 95 patients (475 controls, median age 46 years, 45.3% female). In 2009/10, the number of admissions was lower in IBD patients than in controls (mean 1.53 +/− 1.03 vs. 2.54 +/− 2.35; p < 0.0001). This contrasts with the first audit, where IBD patients had more admissions than controls. Following IBDS introduction, the mean total cost of inpatient care was lower for IBD patients than controls (US$12,857.48 (US$15,236.79) vs. US$30,467.78 (US$ 53,760.20), p = 0.005). In addition, patients known to a specialist gastroenterologist (GE) and the IBD Service tended to have the lowest mean number of admissions (GE and IBDS 1.14 (+/−0.36) vs. no GE/IBDS 1.64 (+/−1.25)).

Conclusions: Healthcare utilisation and disease burden in IBD decreased significantly since introducing an IBDS. These data suggest that proactive management improved outcomes. Contact with a gastroenterologist and IBDS seemed to give best results.

## Introduction

Inflammatory bowel disease (IBD), comprising predominantly ulcerative colitis (UC) and Crohn's disease (CD), is a group of relapsing and remitting inflammatory gastrointestinal disorders. The activity of disease fluctuates over time and therefore, at varying time-points, maintenance therapy and acute interventions to target acute flares of disease activity are both necessary. Whilst maintenance therapy with several agents has been shown to maintain remission (5-aminosalicylic acid (5ASA), azathioprine (AZA), Biologics16), many patients nonetheless suffer disease flares. The reasons for this are likely to be multifactorial and include: impaired access to suitable therapies, intolerance and poor/non-compliance in addition to true disease resistance. In general, whilst some authors propose an aggressive top–down therapeutic approach for patients, the general approach in most centres is to treat patients with the “minimal” effective maintenance therapy needed to prevent or suppress acute flares (e.g. 5-aminosalicylic acid in UC) and then to provide subsequent prompt access to more potent and costly treatment (e.g. biologicals), when required.1

In day-to-day clinical care it is a challenge to provide IBD patients with the level of continuous care required to address their clinical needs, especially if the model used to deliver care is that of an individual doctor treating an individual patient. The provision of good quality care in chronic diseases requires a different approach with greater access to information for patients, education, reminders and empowerment amongst other approaches as discussed by Wagner et al.7 Currently, IBD care is frequently episodic, characterised by health care utilisation when acute flares manifest, and an absence of contact with the treating doctor when the patient is well. IBD healthcare utilisation frequently occurs when complications arise that require hospitalisation8 and overall the treatment of these complications makes IBD costly, both economically and medically. Thus it appears likely that a model of a structured, specialised IBD service might provide a better level of continuous care and improve the outcomes. On the other hand, availability of medically trained IBD specialists is limited and costly. Due to this workforce constraint, a specialised IBD service provided solely by Gastroenterology specialists might not be achievable/affordable for the majority of IBD patients or third party payers. Thus, we hypothesised that a formal IBD service based on a Gastroenterologist-lead shared care model with dedicated, specially trained IBD nurse input would be superior to the standard one-patient–one-doctor model of care as assessed by the need for, and costs of, inpatient health care utilisation. In addition, we aimed to compare the (inpatient) health care utilisation of IBD patients with the (inpatient) health care utilisation of non-IBD patients. We compared data from this current (2009/10) audit with data from the same audit performed in 2007/8, immediately before the introduction of the IBD Service.

## Materials and methods

### Setting and intervention

A prospective cohort study of IBD patients attending a major metropolitan hospital was performed immediately prior to, and again after the introduction of a formal IBD Service within the hospital. Hospital electronic data sources were used to systematically capture all identifiable IBD patients attending the Royal Adelaide Hospital (RAH) over two 5 month periods in 2007/8 (previously published9) and 2009/10 as described below. The RAH is a large metropolitan teaching hospital of 650 beds, serving a local referral population of ~ 600,000 and also providing state-based specialty services for an additional 300–400,000 people.

Prior to April 2008, care for patients with IBD at the RAH was delivered discretely to each patient by individual treating doctors, and not all patients necessarily had specialist gastroenterology care, with 30% of patients with IBD receiving their care within the RAH by non-Gastroenterologists including general physicians, surgeons and emergency room physicians.9,10

In April 2008 a formal IBD Service was introduced for all inpatients and outpatients. This consisted of a named lead clinician, a specialty IBD nurse, a weekly, designated IBD clinic, a joint medical–surgical clinic for IBD and a regular time for radiology review as recommended in the British IBD Standards11. Four experienced gastroenterologists, an IBD fellow and a specialty gastroenterology trainee registrar were regularly involved in delivering IBD care within the Service, although other gastroenterologists continued to also manage their own existing IBD patients and to take on new cases who presented to them during their on-call duty or by specific referral. Thus, not all patients with IBD received care via the IBD Service or indeed via the gastroenterology unit, as this could not be mandated, although it was encouraged. Within the hospital as a whole, the IBD Service spoke at clinical meeting, published and promulgated treatment guidelines and encouraged the referral of IBD patients for assessment/advice. Generally, patients treated via the IBD Service were seen by the same doctor, but cross-covered by another physician when their treating doctor was at a conference, on leave or otherwise absent.

The IBD Service also included:

• Telephone help-line with answers from nurse within 24 h Monday–Friday

• Formal protocols for blood tests monitoring executed by the IBD nurse (no longer handled in an ad hoc fashion or referred to general practitioner (GP)). This meant the nurse would be speaking to all thiopurine patients weekly when starting drug or during dose adjustments, thus ensuring compliance and early detection of adverse drug reactions, and non-response.

• Scheduled phone follow-up for any patient with symptoms at routine clinic visits and post hospital discharge to ensure resolution (pre-empting any deterioration whilst awaiting next routine visit)

• Active nurse management of clinic appointments according to changing patient status (no longer clerically scheduled)

• Single contact point for patient queries, phone & email provided

• Patient education and information leaflets (many specific to RAH clinic protocols) provided for all new patients and all those with treatment changes

• Newsletter on a quarterly basis (postal)

• Encouragement to join Crohn's and Colitis Australia (CCA) (national patient support group).

### Patient cohort

All patients attending the hospital with a confirmed diagnosis of IBD were identified over 5 months from November 1, 2007 to March 31, 2008 via diagnostic and coding data contained in- and outpatient databases as described elsewhere.9

These patients formed the pre-IBD Service whole-of-institution-cohort of IBD patients. From this cohort, all patients with an inpatient admission (defined as one or more overnight stay(s) during a 15 month observation period (January 1, 2007 to March 31, 2008)) were identified. The same audit was duplicated 2 years later (19 months after the introduction of the IBD Service), and again those IBD patients with an inpatient care episode in a matching observation period (January 2009 to March 2010) were identified. Importantly, to avoid bias, the methodology of data collection, length of the observation period and main outcomes were duplicated in both audit periods.

Each admitted IBD case was randomly assigned five age- and gender-matched contemporaneously admitted controls from the whole of hospital coding data (i.e. different time-matched controls for each audit period) using a computer algorithm by the case-mix unit who were not involved in patient care. A large mixed hospital inpatient group was chosen to allow “averaging” of other diagnoses (by using 5 controls per case). The control group was purposely not selected from within the gastroenterology service as other GI conditions are not the appropriate comparator group for IBD as most other admitted GI patients either have acute issues (peptic ulcer, not chronic illness), malignancy or chronic liver disease. The time-matching in the controls provided a benchmark for other temporal trends in the local healthcare environment (such as early discharge pressure).

Data from the initial pre-IBD Service audit has been previously published and showed greater inpatient healthcare utilisation, complexity and costs for admitted IBD patients as compared to controls.9 Here we present the comparison of inpatient healthcare utilisation, care complexity and costings between the audit periods (2007/8 vs. 2009/10), before and after the IBD Service implementation. Specifically we compared outcomes between:

• Discrete IBD patients who were captured in both periods (i.e. those admitted both before and after IBDS started),

• All IBD patients captured in each separate audit, and

• Admitted IBD patients captured in the 2009/10 (post Service introduction) audit compared to their 2009/10 admitted non-IBD controls.

As this was a sample of convenience, and performing the study over a longer period was not possible; as it would not then have been directly comparable to the earlier — already completed audit; a formal statistical power calculation was not performed.

The study was reviewed and approved by the RAH human research ethics committee as clinical audit. As all data was de-identified, and no intervention was administered, specific patient consent was not required.

### IBD patient ascertainment

Three information sources were used to identify IBD patients:

1. Institution-wide computerised databases containing data on all patients with either an inpatient (i.e. admitted overnight or longer) or outpatient encounter, including emergency department visits. For each encounter, diagnoses are recorded using International Classification of Diseases (ICD-10) codes. ICD-10 codes explicitly for IBD and also common IBD-related conditions (e.g. perianal disease, primary sclerosing cholangitis, arthritis, skin and eye conditions) were used.

2. A departmental database (medical and surgical units) containing the vast majority of hospital outpatients' recorded correspondence. Document searches were conducted using keywords “Crohn”, “UC”, “colitis”, “IBD”

3. The endoscopy procedural database. Searches were regularly conducted for patients presenting for colitis surveillance or with newly diagnosed IBD.

IBD patients' case notes, radiology results and pathology were each reviewed to confirm IBD diagnosis and to collect other relevant clinical data. Case notes review was undertaken by a mixture of medical, nursing and research staff according to strict criteria, any uncertainties were adjudicated by the patient-in-question's own treating specialist or by a senior IBD clinician (JMA).

In each of the two audit periods, non-IBD admitted controls were patients identified electronically from discharge data who had been admitted to the same institution during one of the same two 15 month observation periods as their respective IBD cases. To avoid bias by one diagnostic group or another, oversampling and averaging principles were used, by allocating 5 contemporaneously admitted controls to each admitted IBD case. Prior to confirmation as a non-IBD control, ICD-10 coding of primary and secondary diagnoses were re-examined to exclude controls with possible coincident IBD. Controls were age and gender matched to their IBD cases.

### Main outcomes and variables measured

Inpatient healthcare utilisation was measured by:

• the number of inpatient admissions for each patient (index and subsequent re-admissions within the observation period)

• the total length of stay (TLoS, defined as the total, cumulative number of days of all inpatient admissions) of each patient during the observation period and

• the cost of inpatient care (costings of each discrete inpatient admission were provided by the Casemix & Clinical Costing Unit, RAH). Inpatient costing data include allowances for all aspects of hospital care including drugs, pathology, theatre, staff etc. Mean/median cumulative cost of inpatient care (all admissions) per patient and mean/median cost per admission per patient were calculated.

Inpatient healthcare was also assessed for “complexity”. A complex admission being one where the patient is deemed/thought to be sicker than average and therefore expected to consume greater resources. The measurement of complexity of care was done via multiple routine methods:

1. Examining the proportion of admissions resulting from emergency presentations (as opposed to elective admissions).

2. Looking for increased diversity of ICD-10 disease coding categories — a ‘complex’ admission was designated as one where the patient fulfilled three or more unique diagnostic categories during a single admission, implying multiple comorbidities requiring management during the same admission.

3. Using the relative stay index (RSI); defined as the actual length of stay divided by the expected length of stay adjusted for the local hospital's usual casemix. The RSI has been used widely by Australian hospital administrators since the 1970s. A relative stay index (RSI) > 1 indicates a length of stay longer than expected given the hospital's casemix, implying unexpectedly greater complexity for that particular admission.

### Case notes review

Further information was collected from the case notes of IBD patients:

• Personal information (patient record number, name, date of birth, gender)

• Disease specific information (IBD diagnosis, activity of disease during recruitment period, bowel surgery/date of surgery, perianal disease, stoma)

• Healthcare utilisation (date of encounter, admissions during study period)

• Specialist treatment (main IBD treating specialist in the past 12 months, known by a specialist gastroenterologist, known to IBD service)

• Medications (particularly Biologics, Immuno-modulator therapy, steroids, opiates, 5-aminosalicylic acid)

## Analysis

Comparisons of mean number of admissions (includes index admission and any subsequent re-admissions in the study period) and length of stay were conducted using Mann Whitney U tests or Kruskal–Wallis H tests. Chi-square test was used to compare data in form of frequencies in discrete categories. Analyses were 2-tailed and p values < 0.05 were considered statistically significant. For post-hoc analyses, based on the number of comparisons, the Bonferroni adjusted p value is reported.

Statistical analyses were conducted with SPSS 16 (Chicago, IL).

## Results

### First audit compared to second audit for IBD patients captured in both periods (N = 91)

The first audit captured 233 IBD patients, the second audit captured 228. Ninety-one patients were captured in both. In these 91 patients, a greater proportion had an inpatient admission in 2007/8 than in 2009/10 although this was not statistically significant (38.5% vs. 29.7%, see Table 1).

Of note, the mean number of admissions per patient was significantly higher in 2007/8 than in 2009/10 (p = 0.031). There was also a trend for shorter Total length of stay from 2007/8 to 2009/10 (mean 9.90 vs. 5.38, p = 0.072). In addition, many other measures of disease burden were numerically lower in the 2009/10 compared to the 2007/8 audit. The number of patients with increased relative stay index (RSI) (> 100%) was significantly lower in the second audit compared to the first (p = 0.011); and the mean relative stay index (RSI) was significantly lower in 2009/10 than in the 2007/08 (1.07 vs. 1.89, p = 0.02). Although not statistically significant, the proportion of admissions resulting from emergency department (ED) attendances and the number of patients with increased diversity of disease categories for a single admission were also both lower in 2009/10 than in 2007/8.

### First audit compared to second audit for all IBD patients captured in each separate audit (N = 233 cf 228)

Comparing the 233 patients captured in the first audit captured with the second cohort of 228, a small reduction in the proportion of patients admitted was seen (p = 0.0211) (Table 2).

In addition, patients in the second audit had a significantly lower mean number of admissions (1.53 vs. 1.72, p = 0.05), and also, the number of admitted IBD patients with increased relative stay index (> 100%) was very significantly lower in the second compared to the first audit (24 vs. 59 patients, p < 0.0001). Whilst not statistically significant, all other parameters showed a consistent trend to lower inpatient healthcare utilisation in 2009/10 compared to 2007/8: Including Total length of stay (10.52 vs. 13.26, p = 0.348); admissions from ED (77.9% vs. 80.2%, p = 0.692), and increased diversity of disease (46.3% vs. 48.5%, p = 0.758) as shown in Table 2. Importantly, despite biological therapy becoming newly available (yet restricted to patients with Crohn's disease) in Australia between the audit periods (infliximab in October 2007, and adalimumab in August 2008), healthcare utilisation and burden of disease were no different between admitted patients with Crohn's disease (n = 70 in 2007/8 and 60 in 2010) as compared to UC (n = 31 and 25) within or between audit periods (data provided in Table 39).

Patients in the second, as compared to the first, cohort had numerically lower total costs for all admissions (US$1,105,743 vs. US$1,418,581) and lower mean costs per patient (US$12,857.48 +/−$15,236.79 vs. US$14,045.36 +/−$24,509.07; p = 0.545); although these differences did not meet statistical significance.

Disease duration in the admitted patients where data were available was not substantially different amongst audit periods; 9.2 years (available for 85/102 patients; range 0–39 years) in the 2007/8 audit period vs. 11 years (available for 80/95 patients; range 0–50 years) in the 2009/10 period.

### 2009/10 admitted IBD patients compared to age and gender matched controls

63% of admitted patients had Crohn's disease which was no different to the total cohort (64% patients with Crohn's disease). The median age of the whole cohort (n = 228) was 45 years (IQR 31.0, 56.0; age range 67). For the admitted patients (n = 95) median age was 46 years (IQR 36.0, 62.0; age range 66). 55% of admitted patients were males compared to 52% males in the whole cohort. The median age for women (n = 43) was 43 (age rage 66) and for men (n = 52) median age was 51 (age range 65).

Each of these 95 admitted IBD patients was assigned 5 matched controls from all admitted non-IBD patients (n = 41,665) in the same observation period. The total number of admissions for the IBD patients was 145 with a Total length of stay of 999.3 days.

In contrast to our initial audit9 in 2007/8 where IBD patients had significantly greater inpatient healthcare utilisation than controls, the mean number of admissions in 2009/10 was significantly lower for IBD patients than contemporaneous matched admitted controls (1.53 +/− 1.03 vs. 2.54 +/− 2.35, p < 0.0001). Additionally, the mean Total length of stay (LoS) per patient was numerically lower for IBD patients, although this did not reach statistical significance due to wide variation in LoS (10.52 (+/−14.47) vs. 16.58 (+/−26.42), p = 0.46). These trends for decreased inpatient healthcare utilisation in 2009/10 also hold true when CD and UC are considered separately compared to their respective controls (Table 3). In addition, inpatient healthcare utilisation was not different between patients with CD when directly compared to UC (mean admissions 1.4 (0.83) vs. 1.64 (+/−1.08), p = 0.32, median Total length of stay 6.13 (+/−8.67) vs. 6.73 (+/−10.87), p = 0.714).

IBD patients with CD had significantly fewer (re-)admissions than controls (1.4 (SD +/−0.83) vs. 2.66 (SD +/−2.47), p < 0.0001) and although not statistically significant, a numerically lower mean Total length of stay (9.07 (+/−11.95) vs. 16.73 (+/−26.57), p = 0.200). Additionally, IBD patients had a significantly lower mean number of admissions compared to their respective non-IBD controls regardless of whether they had bowel resection prior to the study period or not (previous resection vs. controls: 1.24 (+/−0.51) vs. 2.51 (+/−2.47), p = 0.003) and (no previous resection vs. controls: 1.49 (+/−1.07) vs. 2.53 (+/−2.32), p = 0.003). This finding is in contrast to earlier data showing operated IBD patients to have greater inpatient care needs, prior to the introduction of the IBD Service9.

Although not statistically significant, compared to all other groups, patients who had both specialist gastroenterologist (GE) care and were known to the IBD Service had the lowest mean number of admissions per patient (prior gastroenterologist and IBDS 1.14 (+/−0.36) vs. no gastroenterologist and no IBDS 1.64 (+/−1.25), p = 0.25, Table 4).

The median cumulative length of stay per patient was significantly lower in groups of patients who had specialist gastroenterologist care compared to groups without gastroenterologist care (Table 4). Also, patients who were seen by a gastroenterologist showed a trend to numerically lower median cumulative cost of all admissions and lower median cost per admission per patient compared to those not under gastroenterologist care, although this did not reach statistical significance. This was not explained by any significant difference in complexity between the groups, proportion of Emergency Department admissions or by mean relative stay index (Table 4). Also, other clinical variables such as age, gender and disease activity were not associated with any significant difference between these groups (all p > 0.3). On a quality note, patients with both specialist gastroenterologist care and known to the IBD service did not require any treatment with opiates or other pain killers in contrast to patients in other groups.

### Complexity and costs of inpatient admissions in IBD patients compared to controls

IBD patients had a similar proportion of admissions (index and re-admissions) resulting from emergency presentations and similar diversity of ICD-10 disease categories per admission to controls (as shown below in Table 5).

However, the mean relative stay index was significantly higher in IBD patients than controls (118.3% vs. 113.5%, p = 0.002). Yet, despite this higher relative stay index, the mean total cumulative cost of inpatient care was significantly lower for IBD patients than for respective controls (US$12,857.48 vs. US$30,467.78, p = 0.005, Table 5). The mean cost per admission per patient was also numerically lower for IBD patients, although this difference did not reach statistical significance (US$7946.91 vs. US$11,819.03, p = 0.507).

## Discussion

Our data show, that healthcare utilisation and disease burden in admitted IBD patients had changed significantly during the last two years at our institution. The data suggest that the newly instigated proactive management model with the introduction of an IBD Service has indeed reduced the burden of disease, as assessed by the need for inpatient care in IBD patients (see Fig. 1). The fact that our IBD patients were previously more expensive to care for compared to controls (pre-IBDS) and are now less expensive to care for than controls is strong evidence that something has changed which is specific to the care of IBD patients and not just a change in local resource allocation or healthcare trends. There has been a gradual global trend to deliver IBD care in this way, as promulgated by guidelines,11 however, there has been no previously well-documented clinical or economic benefit shown from adopting this approach. We were able to take advantage of the fact that our institution had not adopted this model prior to 2008, and thus a window of opportunity arose to document the potential benefit of an integrated care model.

It is known that inpatient care is the main driver of direct costs in IBD12 and also that hospital admission correlates with severity of disease,10 and thus is likely to predict poorer outcomes for patients. Thus the changes demonstrated here are likely to be of clinical importance to our patients and also of economic benefit to the healthcare system. Given the short time frame over which the changes are seen, we feel that it is unlikely that other factors have driven the improvements, however one needs to consider potential confounders.

Of note, anti-TNF therapy became available to our patients locally via public funding in August 2007 (infliximab) and April 2008 (adalimumab), and thus one may speculate that these newly available agents substantially contributed to the benefits we demonstrated, given their proven efficacy in reducing hospitalisation.13 In our original audit9 only 10 subjects (of 233) were receiving anti-TNF therapy, whereas in the second period 46/228 were on anti-TNF therapy. However, in Australia, these agents are only available to patients with CD and not UC, yet the gains we demonstrated in decreased healthcare utilisation were seen equally in patients with CD and UC. Furthermore only a small proportion (11/95, Table 4) of the admitted patients captured in the second audit period were receiving anti-TNF therapy and we thus feel it is unlikely to be the main driver of decreased healthcare utilisation, at least in admitted patients, although we would speculate (and hope) that it may have averted the need for other admissions in patients with CD.

It is also possible that a greater and earlier use of other immunosuppressants may have contributed to the benefits we found. However, in our initial cohort study the rate of use of thiopurines was already 42% overall10 (~ 20% in patients with UC and ~ 60% in those with CD), making it unlikely that more aggressive thiopurines use alone could have resulted in these improvements. However, on the other hand, monitoring of thiopurine compliance and dosing by the IBD nurse with reminder calls for safety blood tests may well have improved compliance and enhanced efficacy of these agents where they were already prescribed, but not previously taken as directed. We cannot dismiss whether combination treatment with anti-TNF and immunosuppressants may contribute to reduced health care utilisation between the two cohorts. In the first cohort, almost no Biologics were used, so neither was combination therapy. In the second admitted cohort, only 11 patients with Crohn's disease were on biologics with most also on an immunosuppressant (no UC patient had biologics). As these numbers are so small no formal comparison of outcomes with respect to combination therapy can be made, but as the same trends are seen for patients with UC (without combination therapy), this is unlikely to be driving the change in healthcare utilisation.

As IBD is a chronic disease and it is difficult to tailor care contemporaneously to patients' needs when they are only seen episodically, it is highly likely that methods (such as ours) which enhance their access to timely clinical advice/contact (and include more routine contacts by nurse, phone, email) will improve clinical outcomes and reduce the need for inpatient care. This approach of enhancing patient care by introducing extended nursing roles has been previously validated in many other disease states. In particular nurse-led telephone follow-up systems have been shown to improve outcomes of care in patients with hypertension, diabetes and brain injuries.1418 Reductions in the number of readmissions and improvements regarding compliance are widely reported after the introduction of simple interventions such as the introduction of telephone monitoring.15,17,19

A systematic review regarding the potential of telephone follow-up for patients with colorectal cancer showed that telephone care is cost-effective and accepted by the majority of patients.20 The authors claim that telephone consultations are at least equivalent to traditional follow-up in meeting the needs of patients with cancer. Further studies showed that patients appreciate the nurse-initiated telephone calls and use these calls to get information, referrals and emotional support.21,22 Also, a study published recently showed that different models of care obviously have the potential to deliver different outcomes.23

Whilst employing the IBD nurse is an additional upfront cost in the provision of this service, there was actually a net cost saving in this approach. When one considers the mean costs per admission in our latter audit was US$7946.91 (Table 5), and that 30 admissions were saved (Table 2), there is a direct cost saving of ~ US$238,407 when just considering the patients contacting the hospital in a five-month period. If one were to continue the audit for a full year, the direct savings would be even higher, and are clearly far greater than the annual costs of a fulltime nurse (~ US$80,000), or two. It is also important to emphasise that these are only the direct savings and do not take into account, reduced patient morbidity/disability or loss of earnings etc. Whilst there are many different payment systems for healthcare, we believe that our results are generalisable. Our local model for healthcare delivery actually makes it easier to capture data on a higher proportion of patients with a single disease living in one area, and so more complete than in locations where patient may choose or be directed to use a variety of facilities. None-the-less, regardless of where care is delivered, if fewer patients have complex admissions and their length of stay is decreased with reduced inpatient costs, there will be a benefit in any system and a likely reduced morbidity to patients. In the US, for example, where costs for healthcare are much higher than in Australia, so the gains are also likely to be greater. Some limitations need to be acknowledged: First, a pre–post study design may have some potential for bias. Yet it is the only practical method by which this data (and this change in care model) can be tested in practical terms. For example, applying different models of care contemporaneously in different centers would involve far more biases. Additionally, we controlled for temporal bias by the use of contemporaneous controls. By the use of two different contemporaneous control groups we cancel out the issue of “institutional” factors (i.e. more efficient and aggressive discharging) over time as any factors should have affected IBD patients and controls in the same fashion. Second, the control group consists of patients with disorders other than IBD. However, this is the best control group as each admitted IBD case was randomly assigned five age- and gender-matched contemporaneously admitted controls to overcome bias from any particular other condition. Non-admitted IBD patients would not be an appropriate control group and there is no particular accepted disease control which is perfect for IBD. Third, we focus on health economics/costs rather than clinical outcomes such as morbidity, however reducing costs and improving efficiencies in healthcare are vital, as this allows us to deliver better care to more patients. Finally studying outcomes solely in admitted patients is just the tip of the iceberg as we considered only patients who were acutely ill and warrant hospitalisation. But other data are difficult to gather as completely as data from inpatients case notes as one cannot compel un-admitted patients (who may not even be attending the hospital regularly if well — perhaps annually or less) to provide clinical details. Additionally, data on ambulatory patients can suffer from response bias. One of the strengths of this approach is that there is no responder bias possible. In summary, we have clearly demonstrated financially and clinically relevant gains for IBD patients by formally adopting a chronic care model with the assistance of a dedicated IBD nurse coordinating care, especially pro-active, out-of-hospital contact. This approach, whilst necessitating increased staffing, is cost-effective leading to net savings in direct healthcare costs. The indirect gains to patients quality of life are likely to be greater than demonstrated here, and we would now recommend this care model as best-practice, as it appears in our experience to be effective at reducing admissions, and also reducing the costs and duration of each admission for IBD. However, a word of caution is necessary, as we have not directly assessed outpatient costs for care, medications, radiology and outpatient office visits. These items and clear clinical outcomes should be measured in future studies. ## Competing interests None to declare. ## Funding No funding or writing assistance was received for this study. ## Conflict of interest statement None to declare. ## Acknowledgments We would like to acknowledge funding for the initial two years of our extended roles IBD nurse from pharmaceutical companies involved in IBD-related products including: Schering-Plough (infliximab), Abbott (adalimumab), Ferring (pentasa), Orphan (salofalk) and Fresenius Kabi (colazide). Our two IBD specialist nurses are now funded by the RAH nursing budget. ## References 1 D'Haens G. Baert F. van Assche G. Caenepeel P. Vergauwe P. Tuynman H. De Vos M. van Deventer S. Stitt L. Donner A. Vermeire S. Van de Mierop F.J. Coche J.C. van der Woude J. Ochsenkuhn T. van Bodegraven A.A. van Hootegem P.P. Lambretcht G.L. Mana F. Rutgeerts P. Feagan B.G. Hommes D. Early combined immunosuppression or conventional management in patients with newly diagnosed Crohn's disease: an open randomised trial Lancet 371 2008 660 667 2 Fraser A.G. Orchard T.R. Jewell D.P. The efficacy of azathioprine for the treatment of inflammatory bowel disease: a 30 year review Gut 50 4 2002 485 489 3 Bebb J.R. Scott B.B. How effective are the usual treatments for ulcerative colitis? Aliment Pharmacol Ther 20 2 2004 143 149 4 Sutherland L. Macdonald J.K. Oral 5-aminosalicylic acid for maintenance of remission in ulcerative colitis Cochrane Database Syst Rev 2 2006 CD000544 5 Colombel J.F. Sandborn W.J. Rutgeerts P. Enns R. Hanauer S.B. Panaccione R. Schreiber S. Byczkowski D. Li J. Kent J.D. Pollack P.F. Adalimumab for maintenance of clinical response and remission in patients with Crohn's disease: the CHARM trial Gastroenterology 132 1 2007 52 65 6 Hanauer S.B. Feagan B.G. Lichtenstein G.R. Mayer L.F. Schreiber S. Colombel J.F. Rachmilewitz D. Wolf D.C. Olson A. Bao W. Rutgeerts P. Maintenance infliximab for Crohn's disease: the ACCENT I randomised trial Lancet 359 9317 2002 1541 1549 7 Wagner E.H. Austin B.T. Von Korff M. Organizing care for patients with chronic illness Milbank Q 74 4 1996 511 544 8 Blomqvist P. Feltelius N. Lofberg R. Ekbom A. A 10-year survey of inflammatory bowel diseases-drug therapy, costs and adverse reactions Aliment Pharmacol Ther 15 4 2001 475 481 9 Langenberg DRv Simon S.B. Holtmann G.J. Andrews J.M. The burden of inpatient costs in inflammatory bowel disease and opportunities to optimize care: a single metropolitan Australian center experience J Crohns Colitis 4 4 2010 413 421 10 van Langenberg D.R. Lange K. Hetzel D.J. Holtman G.J. Andrews J.M. Adverse clinical phenotype in inflammatory bowel disease: a cross sectional study identifying factors potentially amenable to change J Gastroenterol Hepatol 25 7 2010 1250 1258 11 British IBD Standards Oct. 2010 12 Blomqvist P. Ekbom A. Inflammatory bowel diseases: health care and costs in Sweden in 1994 Scand J Gastroenterol 32 11 1997 1134 1139 13 Lichtenstein G.R. Yan S. Bala M. Blank M. Sands B.E. Infliximab maintenance treatment reduces hospitalizations, surgeries, and procedures in fistulizing Crohn's disease Gastroenterology 128 4 2005 862 869 14 Rudd P. Miller N.H. Kaufman J. Kraemer H.C. Bandura A. Greenwald G. Debusk R.F. Nurse management for hypertension. A systems approach Am J Hypertens 17 10 2004 921 927 15 Wasson J. Gaudette C. Whaley F. Sauvigne A. Baribeau P. welch H.G. Telephone care as a substitute for routine clinic follow-up JAMA 267 13 1992 1788 1793 16 Piette J.D. Weinberger M. McPhee S.J. The effect of automated calls with telephone nurse follow-up on patient-centered outcomes of diabetes care: a randomized, controlled trial Med Care 38 2 2000 218 230 17 Piette J.D. Weinberger M. McPhee S.J. Mah C.A. Kraemer F.B. Crapo L.M. Do automated calls with nurse follow-up improve self-care and glycemic control among vulnerable patients with diabetes? Am J Med 108 1 2000 20 27 18 Bell K.R. Temkin N.R. Esselman P.C. Doctor J.N. Bombardier C.H. Fraser R.T. Hoffman J.M. Powell J.M. Dikmen S. The effect of a scheduled telephone intervention on outcome after moderate to severe traumatic brain injury: a randomized trial Arch Phys Med Rehabil 86 5 2005 851 856 19 Madge P. McColl J. Paton J. Impact of a nurse-led home management training programme in children admitted to hospital with acute asthma: a randomised controlled study Thorax 52 3 1997 223 228 20 Cusack M. Taylor C. A literature review of the potential of telephone follow-up in colorectal cancer J Clin Nurs 19 17–18 2010 2394 2405 21 Anderson B. The benefits to nurse-led telephone follow-up for prostate cancer Br J Nurs 19 17 2010 1085 1090 22 Keeling A.W. Dennison P.D. Nurse-initiated telephone follow-up after acute myocardial infarction: a pilot study Heart Lung 24 1 1995 45 49 23 Magro F. Barreiro-de Acosta M. Lago P. Carpio D. Cotter J. Echarri A. Goncalves R. Pereira S. Carvalho L. Lorenzo A. Barros L. Castro J. Dias J.A. Rodrigues S. Portela F. Dias C. da Costa-Pereira A. Clinical practice in Crohn's disease in bordering regions of two countries: different medical options distinct surgical events. J Crohns Colitis 4 2010 301 311 ## Figures Figure 1 Proposed newer care model. Figure 1 Proposed newer care model. ## Tables Table 1 Inpatient healthcare utilisation and disease burden for patients captured in both periods (2007/8 and 2009/10) (N = 91).  First cohort 2008 Second cohort 2010 p value Patients with inpatient admissions (in % of similar cases) 35 (38.5%) 27 (29.7%) 0.211 Sum of inpatient admissions 62 34 Mean number of admissionsa per patient (SDc) 1.77 (1.06) 1.26 (0.66) 0.031 Mean cumulative total length of stay per patient (SDc) 9.90 (15.37) 5.38 (6.73) 0.072 Admissions resulting from EDb 28 (80%) 19 (70.4%) 0.38 Increased diversity of disease categories within single admissionsb 12 (34.29%) 8 (29.63%) 0.697 Mean relative stay index (RSI)b 1.89 (2.60) 1.07 (0.99) 0.02 Relative stay index (RSI) > 100%b 19 (54.3%) 6 (22.2%) 0.011  First cohort 2008 Second cohort 2010 p value Patients with inpatient admissions (in % of similar cases) 35 (38.5%) 27 (29.7%) 0.211 Sum of inpatient admissions 62 34 Mean number of admissionsa per patient (SDc) 1.77 (1.06) 1.26 (0.66) 0.031 Mean cumulative total length of stay per patient (SDc) 9.90 (15.37) 5.38 (6.73) 0.072 Admissions resulting from EDb 28 (80%) 19 (70.4%) 0.38 Increased diversity of disease categories within single admissionsb 12 (34.29%) 8 (29.63%) 0.697 Mean relative stay index (RSI)b 1.89 (2.60) 1.07 (0.99) 0.02 Relative stay index (RSI) > 100%b 19 (54.3%) 6 (22.2%) 0.011 a Including index admission and subsequent re-admission(s) in observation period. b Based on the index admission. c Standard deviation. Table 2 Inpatient healthcare utilisation and disease burden in admitted patients captured in each separate audit.  First audit 2008 (n = 233) Second audit 2010 (n = 228) p value Patients with inpatient admissions 102/233 95/228 Proportion of patients in the cohort requiring admission 43.8% 41.7% 0.0211 Sum of inpatient admissions 175 145 Mean number of admissionsa per patient (SDc) 1.72 (0.97) 1.53 (1.03) 0.05 Mean cumulative total length of stay per patient (SDc) 13.26 (21.11) 10.52 (14.47) 0.348 Admissions resulting from EDb 81 74 0.692 Increased diversity of disease categories within single admissionsb 49 44 0.758 Mean relative stay index (RSI)b 1.99 (2.33) 1.04 (0.73) < 0.0001 Relative stay index (RSI) > 100%b 59 24 < 0.0001  First audit 2008 (n = 233) Second audit 2010 (n = 228) p value Patients with inpatient admissions 102/233 95/228 Proportion of patients in the cohort requiring admission 43.8% 41.7% 0.0211 Sum of inpatient admissions 175 145 Mean number of admissionsa per patient (SDc) 1.72 (0.97) 1.53 (1.03) 0.05 Mean cumulative total length of stay per patient (SDc) 13.26 (21.11) 10.52 (14.47) 0.348 Admissions resulting from EDb 81 74 0.692 Increased diversity of disease categories within single admissionsb 49 44 0.758 Mean relative stay index (RSI)b 1.99 (2.33) 1.04 (0.73) < 0.0001 Relative stay index (RSI) > 100%b 59 24 < 0.0001 a Including index admission and subsequent re-admission(s) in observation period. b Based on the index admission. c Standard deviation. Table 3 Healthcare utilisation (HCU) in IBD subgroups and respective matched controls (2009/10 audit period).  IBD subgroup Variable IBD subgroup Controls p value Crohn's disease N 60 300 Mean admissionsa (SD) 1.4 (0.83) 2.66 (2.47) < 0.0001 Median TLoS (IQRb) 6.13 (8.67) 6.68 (17.47) 0.2000 Mean TLoS (SDb) 9.07 (11.95) 16.73 (26.57) Ulcerative colitis N 25 125 Mean admissions (SDb) 1.64 (1.08) 2.50 (2.15) 0.122 Median TLoS (IQRb) 6.73 (10.87) 6.81 (20.79) 0.5740 Mean TLoS (SDb) 11.56 (17.72) 17.42 (25.39) Indeterminate N 10 50 Mean admissions (SDb) 2.00 (1.76) 1.94 (2.02) 0.8420 Median TLoS (IQRb) 10.89 (22.33) 3.53 (8.15) 0.0710 Mean TLoS (SDb) 16.59 (18.92) 13.25 (28.24) IBD, bowel resection surgery during study period (01/2009–03/2010) N 15 75 Mean admissions (SDb) 2.13 (1.36) 2.81 (2.44) 0.4910 Median TLoS (IQRb) 11.78 (28.75) 7.73 (21.08) 0.2270 Mean TLoS (SDb) 22.25 (25.10) 18.84 (27.28) IBD, bowel resection in past (before 01/2009) N 29 145 Mean admissions (SDb) 1.24 (0.51) 2.51 (2.47) 0.0030 Median TLoS (IQRb) 6.75 (6.84) 6.48 (17.46) 0.6050 Mean TLoS (SDb) 7.68 (7.29) 15.01 (21.16) IBD, no history of surgery N 53 265 Mean admissions (SDb) 1.49 (1.07) 2.53 (2.32) 0.0030 Median TLoS (IQRb) 4.76 (9.38) 5.62 (17.55) 0.1350 Mean TLoS (SDb) 8.49 (11.58) 17.19 (29.29)  IBD subgroup Variable IBD subgroup Controls p value Crohn's disease N 60 300 Mean admissionsa (SD) 1.4 (0.83) 2.66 (2.47) < 0.0001 Median TLoS (IQRb) 6.13 (8.67) 6.68 (17.47) 0.2000 Mean TLoS (SDb) 9.07 (11.95) 16.73 (26.57) Ulcerative colitis N 25 125 Mean admissions (SDb) 1.64 (1.08) 2.50 (2.15) 0.122 Median TLoS (IQRb) 6.73 (10.87) 6.81 (20.79) 0.5740 Mean TLoS (SDb) 11.56 (17.72) 17.42 (25.39) Indeterminate N 10 50 Mean admissions (SDb) 2.00 (1.76) 1.94 (2.02) 0.8420 Median TLoS (IQRb) 10.89 (22.33) 3.53 (8.15) 0.0710 Mean TLoS (SDb) 16.59 (18.92) 13.25 (28.24) IBD, bowel resection surgery during study period (01/2009–03/2010) N 15 75 Mean admissions (SDb) 2.13 (1.36) 2.81 (2.44) 0.4910 Median TLoS (IQRb) 11.78 (28.75) 7.73 (21.08) 0.2270 Mean TLoS (SDb) 22.25 (25.10) 18.84 (27.28) IBD, bowel resection in past (before 01/2009) N 29 145 Mean admissions (SDb) 1.24 (0.51) 2.51 (2.47) 0.0030 Median TLoS (IQRb) 6.75 (6.84) 6.48 (17.46) 0.6050 Mean TLoS (SDb) 7.68 (7.29) 15.01 (21.16) IBD, no history of surgery N 53 265 Mean admissions (SDb) 1.49 (1.07) 2.53 (2.32) 0.0030 Median TLoS (IQRb) 4.76 (9.38) 5.62 (17.55) 0.1350 Mean TLoS (SDb) 8.49 (11.58) 17.19 (29.29) a Counting both index and subsequent re-admissions per patient in observation period. b SD = standard deviation, IQR = interquartile range. Table 4 Healthcare utilisation and cost measures in IBD subgroups (2009/10 audit period).  No GEa and no IBDS No GEa and IBDS GEa and no IBDS GEa and IBDS p value Total number of patients with admission 25 14 42 14 Mean number of admissionsb per patient (SDc) 1.64 (1.25) 1.93 (1.27) 1.45 (0.92) 1.14 (0.36) 0.188 Median cumulative length of stay per patient (days) (IQRc) 7.29 (7.60) 8.35 (30.81) 4.83 (10.71) 5.87 (4.92) 0.05 Admissions resulting from ED presentation (in% of patients in this group) 24 (96%) 13 (93%) 32 (76.2%) 10 (71.4%) 0.082 Admissions with increased diversity of disease (in% of patients in this group) 14 (56%) 8 (57.1%) 19 (45.2%) 6 (42.9%) 0.729 Mean relative stay index (RSI) (SDc) 0.98 (0.53) 1.02 (0.70) 1.11 (0.91) 1.00 (0.39) 0.825 No of patients on biologics during study period 0 (0%) 1 (7.14%) 10 (23.81%) 0 (0%) No of patients on I/S during study period 6 (24%) 7 (50%) 24 (57.14%) 8 (57.14%) 0.054 No of patients on steroids during study period 8 (32%) 6 (42.86%) 14 (33.33%) 10 (71.43%) 0.064 No of patients on opiates/pain killers during study period 2 (8%) 4 (28.57%) 13 (30.95%) 0 (0%) No of patients on 5ASA during study period 7 (28%) 4 (28.57%) 21 (50%) 8 (57.14%) 0.141 Median cumulative cost of all admissions in observation period (per patient, US$d) (IQRd) 10,565.50 8,552.00 8,021.50 6,001.00 0.209 Median cost per admission (per patient, US$d) in observation (IQRc) 8,362.83 7,512.33 4,824.25 4,737.00 0.469  No GEa and no IBDS No GEa and IBDS GEa and no IBDS GEa and IBDS p value Total number of patients with admission 25 14 42 14 Mean number of admissionsb per patient (SDc) 1.64 (1.25) 1.93 (1.27) 1.45 (0.92) 1.14 (0.36) 0.188 Median cumulative length of stay per patient (days) (IQRc) 7.29 (7.60) 8.35 (30.81) 4.83 (10.71) 5.87 (4.92) 0.05 Admissions resulting from ED presentation (in% of patients in this group) 24 (96%) 13 (93%) 32 (76.2%) 10 (71.4%) 0.082 Admissions with increased diversity of disease (in% of patients in this group) 14 (56%) 8 (57.1%) 19 (45.2%) 6 (42.9%) 0.729 Mean relative stay index (RSI) (SDc) 0.98 (0.53) 1.02 (0.70) 1.11 (0.91) 1.00 (0.39) 0.825 No of patients on biologics during study period 0 (0%) 1 (7.14%) 10 (23.81%) 0 (0%) No of patients on I/S during study period 6 (24%) 7 (50%) 24 (57.14%) 8 (57.14%) 0.054 No of patients on steroids during study period 8 (32%) 6 (42.86%) 14 (33.33%) 10 (71.43%) 0.064 No of patients on opiates/pain killers during study period 2 (8%) 4 (28.57%) 13 (30.95%) 0 (0%) No of patients on 5ASA during study period 7 (28%) 4 (28.57%) 21 (50%) 8 (57.14%) 0.141 Median cumulative cost of all admissions in observation period (per patient, US$d) (IQRd) 10,565.50 8,552.00 8,021.50 6,001.00 0.209 Median cost per admission (per patient, US$d) in observation (IQRc) 8,362.83 7,512.33 4,824.25 4,737.00 0.469 a GE = gastroenterologist. b Counting both index and subsequent re-admissions per patient in observation period. c SD = standard deviation, IQR = interquartile range. d US$ = United States Dollar.

Table 5

Complexity/cost measures for IBD patients and matched controls (2009/10 audit period).

 Complexity variable IBD patients Controls p value Admissionsa resulting from emergency presentations 105 (72.4%) 922 (76.7%) 0.251 Those with increased diversity of disease categories within single admission 66 (45.5%) 586 (48.8%) 0.462 Relative stay index (RSI) > 100% 48 (33.1%) 356 (29.6%) 0.387 Mean relative stay index (RSI) (SD (%)) 118.3% (87.53%) 113.52% (157.29%) 0.002 Median cumulative cost of all admissions in observation period (per patient, US$) (IQR)$8,563.50 ($12,819.75)$11,683.00$($30,367.00) 0.005 Mean cumulative cost of all admissions in observation period (per patient, US$) (SD)$12,857.48 ($15,236.79)$30,467.78 ($53,760.20) Median cost per admission (per patient, US$) in observation (IQR) $6,388,00 ($8,015.50) $5,950.00 ($11,505.20) 0.507 Mean cost per admission (per patient, US$) in observation (IQR)$7,946.91 ($6,228.51)$11,819.03 ($17,042.33) 0.507  Complexity variable IBD patients Controls p value Admissionsa resulting from emergency presentations 105 (72.4%) 922 (76.7%) 0.251 Those with increased diversity of disease categories within single admission 66 (45.5%) 586 (48.8%) 0.462 Relative stay index (RSI) > 100% 48 (33.1%) 356 (29.6%) 0.387 Mean relative stay index (RSI) (SD (%)) 118.3% (87.53%) 113.52% (157.29%) 0.002 Median cumulative cost of all admissions in observation period (per patient, US$) (IQR) $8,563.50 ($12,819.75) $11,683.00$ ($30,367.00) 0.005 Mean cumulative cost of all admissions in observation period (per patient, US$) (SD) $12,857.48 ($15,236.79) $30,467.78 ($53,760.20) Median cost per admission (per patient, US$) in observation (IQR)$6,388,00 ($8,015.50)$5,950.00 ($11,505.20) 0.507 Mean cost per admission (per patient, US$) in observation (IQR) $7,946.91 ($6,228.51) $11,819.03 ($17,042.33) 0.507
a

Counting both index and subsequent re-admissions per patient in observation period.