Published online Nov 6, 2018. doi: 10.12998/wjcc.v6.i13.624
Peer-review started: July 16, 2018
First decision: September 10, 2018
Revised: September 16, 2018
Accepted: October 11, 2018
Article in press: October 11, 2018
Published online: November 6, 2018
There has been in an increase in the frequency of incidental pancreatic cyst diagnosis. The management of these pancreatic cysts is a significant challenge that clinicians have to face because while the overall risk of malignancy in incidentally detected pancreatic cysts is low, pancreatic cancer has an extremely poor prognosis once acquired. Various gastroenterology societies have published pancreatic cyst management guidelines, one of which was the Sendai guidelines established by the International Association of Pancreatology (IAP). IAP updated their guidelines in 2012, known as the Fukuoka guidelines, in which the resection criteria was modified to be more conservative. Based on the Fukuoka guidelines, Kaiser Permanente Southern California (KPSC) published a regional algorithm in October 2013 with radiology reports including a concise summary of recommendations.
Healthcare budget is limited in all countries around the world. Therefore, there is increased pressure to optimize resource utilization. We wanted to demonstrate through our study that the wide dissemination of an algorithm for pancreatic cyst management in an integrated health system such as KPSC can decrease the endoscopic ultrasound/gastroenterology/surgery referrals from the frontline physicians who detect incidental pancreatic cysts on cross sectional imaging, which then would result in significant cost savings.
The objective of this study was to examine the practice pattern in KPSC, i.e., gastroenterology (GI)/surgery referrals and endoscopic ultrasound (EUS), for pancreatic cystic neoplasms (PCN) after the region-wide dissemination of the pancreatic cyst management algorithm derived from the Fukuoka guidelines published in 2012.
This was a retrospective analysis of patients who were given the PCN diagnosis between April 2012 and April 2015 (18 mo before and after the publication of the algorithm) in KPSC (integrated health system with 15 hospitals and 202 medical offices in Southern California). Patients over the age of 18 years with a diagnosis of pancreatic cysts based on International Classification of Disease - Clinical Modification 9 code (ICD-9 577.2) were included. Diagnosis of pancreatic malignancy was confirmed through cross-reference with a prospectively maintained internal KPSC cancer registry. For cost analysis portion of the study, the cost of EUS was based on 2017 Medicare national average payment for hospital out-patient procedure (CPT® 43238). For GI and surgery consultations, CPT® codes 99201-99205 were used. Extrapolation of the result to the US population was based on the estimation of the incidence of pancreatic cyst in the US being between 3% and 15%, percentage increasing with age, according to the American Gastroenterological Association Technical Review on the Diagnosis and Management of Asymptomatic Neoplastic Pancreatic Cysts. Lower end of this range (3%) was used in our analysis to err on the side of a conservative estimate.
2558 (1157 pre- and 1401 post-algorithm) received a new diagnosis of PCN in the study period. A smaller percentage of PCNs resulted in EUS after the implementation of the algorithm (pre- 45.5% vs post- 34.8%, P < 0.001). A smaller proportion of patients were referred for GI (pre- 65.2% vs post- 53.3%, P < 0.001) and surgery consultations (pre- 24.8% vs post- 16%, P < 0.001) for PCN after the implementation. The incidence rate of pancreatic cancer after diagnosis of PCN (between the time of PCN diagnosis and May 1, 2017 or end of KPSC membership), excluding those that were diagnosed at the time of initial work-up, was 2.41 per 1000-person years (95%CI: 1.16-4.44) for pre-algorithm and 2.42 per 1000-person years (95%CI: 1.04-4.77) for post-algorithm (P = 0.4943). After the implementation, cost of diagnostic care was reduced by 24%, 18%, and 36% for EUS, GI, and surgery consultations, respectively. Total cost saving was 24%. If we assume that the incidence of pancreatic cysts in KPSC is 1000 (estimate based on our search using ICD-9), total annual cost saving would be $194696 ($584088 during the entire study period), $173874 for EUS, $9805 for GI consultation, and $11018 for surgery consultation. However, this incidence may be significantly underestimated; KPSC incidence using the lower end of the US incidence (3%, range 3%-15%) would be 98040 given that KPSC adult population is 3268000. Using this figure as the incidence, total cost saving would increase to $19 Million ($17 Million for EUS, $1 Million for GI consultations, and $1 Million for surgery consultations). These findings were then extrapolated to the national level. The US population estimate as of July 1st, 2017 was 325719178 with 251455205 adults. Using the 3% incidence rate, the cost saving as a result of such algorithm implementation would be $1.49 Billion ($1.34 Billion for EUS, $72 Million for GI consultations, and $81Million for surgery consultations).
Our study is unique in that it examines that economic effect of adopting a pancreatic cyst management guideline in an effective way within a healthcare system. Although there have been numerous studies looking at the clinical effect of the guidelines, there has been a paucity of studies dealing with the change in the practice/referral pattern. Our study demonstrated that the dissemination of an algorithm for pancreatic cyst management in such a manner increased the threshold for referrals for intervention and workup, possibly due to increasing the confidence level of physicians ordering imaging studies in managing PCNs, resulting in significant cost savings without compromising patient care. As stated above, if the findings of this study were extrapolated to the US national level, the cost saving as a result of such algorithm implementation could be $1.49 Billion. There would be a much stronger impact if the recommendations were applied to clinical practice worldwide.
Although our study presents a strong economic message in a relatively short study period of 36 mo and proves that there was no difference in the pancreatic cancer rate between the two groups after following the patients for a longer period until May 1, 2017 (60 mo from the beginning of the study period), we realize that this is not a long enough follow up timeframe to examine the natural course of PCNs. While the aim of the study was to show the cost savings that could result from the change in clinical practice following the dissemination of such an algorithm and not necessarily to show how the algorithm can improve patient outcome, we do realize the importance of studies that address the long-term outcome of not only the PCN algorithm(s) but also the natural course of PCNs themselves. This type of long-term study should be the direction of future research of PCNs.