Decrease Hospital Spending: There's an App for That! A Retrospective Analysis of Implementation of a Mobile Resident Handbook on Hospital Costs and Disposition

Abstract

Background:

Patient care involves time sensitive decisions. Matching a patient's presenting condition with possible diagnoses requires proper assessment and diagnostic tests. Timely access to necessary information leads to improved patient care, better outcomes, and decreased costs.

Introduction:

This study evaluated objective outcomes of the implementation of a novel Resident Handbook Application (RHAP) for smart phones.

Methods:

The RHAP included tools necessary to make proper assessments and to order appropriate tests. The RHAPs effectiveness was accessed using the Military Health System Military Mart database. This database includes patient specific aggregate data, including diagnosis, patient demographics, itemized cost, hospital days, and disposition status. Multivariable analysis was used to compare before and after RHAP implementation, controlling for patient demographics and diagnosis. Internal medicine admission data were used as a control group.

Results:

There was a statistically significant decrease in laboratory costs and a strong trend toward statistically significant decreases in the cost of radiology performed after implementation of RHAP (p value of <0.02 and <0.07, respectively). There was also a decrease in hospital days (3.66–3.30 days), in total cost per admission ($18,866–$16,305), and in cost per hospital day per patient ($5,140–$4,936). During the same time period a Control group had no change or increases in these areas.

Conclusions:

The use of the RHAP resulted in decreases in costs in a variety of areas and a decrease in hospital bed days without any apparent negative effect upon patient outcomes or disposition status.

Introduction

Mobile technology is almost everywhere in today's society. As of 2012, 65% of people in the United States had a smart device (e.g., cellular phone, tablet, etc.), actually carrying one or more devices with them at all times.¹ Mobile applications (Apps) are now the preferred method of obtaining information.² Apps were originally developed for quick access to personal data such as e-mail, calendars, and schedules. They have been incorporated in many aspects of daily and professional life, including news, communication, education, and recreation.^3,4

A study completed in 2011 found that 88.4% of medical residents used a smartphone for medical purposes, most commonly for clinical practice.⁵ It would seem a safe guess that current usage is even higher. A 2012 study found that medical students and residents reported that the most common medical uses of smart phones were accessing medical reference applications (46%), accessing e-books (45%), and studying for medical board certification (32%).⁶

Residents at most hospitals are given physical “Handbooks” with essential contact information, hospital specific procedural algorithms, and evidence-based medical information. These “Handbooks” are helpful in providing quick access to vital information.^7,8 A recent study by the author involved the design of a neurology resident mobile application, the Resident Handbook Application (RHAP). This App focused on the most commonly used and requested information. It was introduced in May 2015 and fully implemented in June 2015 at Walter Reed National Military Medical Center. This mobile App gave residents immediate access to department contact information, acute care management guidelines, assistance with basic neurologic workups, resident schedules, and board review materials. Resident surveys conducted pre- and post-RHAP implementation found statistically significant improvements in residents' abilities to access diagnostic algorithms and essential department specific information. Residents also reported that patient care was improved with the use of RHAP. This is a follow-up study to determine if objective data exist to confirm resident impressions about patient care being improved and if other benefits exist with the use of RHAP.

Methods

The RHAP application was constructed using (1) Information from the hard copy resident handbook, (2) resident schedules, (3) staff rosters, (4) The Emergency Neurologic Life Support (ENLS) website, and (5) The American Academy of Neurology (AAN) treatment guidelines. Information was either entered in PDF format or was accessible with embedded hyperlinks. The RHAP was constructed using the Appmakr © Online interface with no fees to create, download, or to use the App. It was compatible with all major cellular telephones and other “smart” accessories. All residents reported they possessed a smart phone capable of downloading the App. The App required wireless Internet that was available from all carriers and free throughout the Medical Center. A total of three 10-min tutorials were provided to the Neurology Department residents on use of the App with all residents attending a session.

The home screen of the RHAP had multiple tabs; the “Scutbook” tab contained (1) the Table of Contents, (2) Guidelines for managing common neurologic diseases/emergencies, and (3) Admission and treatment logistic information. The “Contacts” tab contained the Neurology Department contact roster (a one-touch system). The ENLS tab contained a link to that website (no cost). The “Schedule” tab provided the most up-to-date Neurology resident on-call schedule with staff backups. The “Guidelines” tab contained up-to-date management guidelines for nonacute neurologic conditions from the AAN guidelines website. Various other Internet links and resources appropriate to patient care were also included. No copyrighted or licensed material was copied or distributed outside of the Neurology Department as a part of this study or in the development of the RHAP application.

The author designed a survey, which was completed by the residents 3 months post-RHAP implementation. Every question was answered using a three-point 2-tailed Likert scale of “Agree,” “Disagree,” or “Neutral” in response to a statement of the RHAPs positive effect on an aspect of Neurology practice. There were four categories as follows: (1) Access to important administrative information, (2) Time critical and commonly used patient management guidelines, (3) RHAP effects upon resident study experiences and access to study resources, and (4) General satisfaction questions to assess the effect RHAP had on patient care.

The Military Health System (MHS) Military Mart (M2) health records database was used to collect objective data before and after RHAP implementation. The Neurology inpatient service (study group) and Internal Medicine (IM) inpatient service (control group) were reviewed from May 2014 through June 2016, 13 months pre- and post-RHAP. Cutoff between the two periods was June 1, 2015. The following data were collected: (1) Total number of admissions, (2) Total number of hospital bed days, (3) Full cost of admissions, using M2 database calculations, and (4) Patient disposition status. Mean bed days per patient and mean cost per patient bed day were calculated. Multivariable data analysis software was used to assess statistical significance before and after RHAP implementation. Independent variables, including patient demographics and primary diagnoses, were controlled for when looking at difference in the pre- and post-RHAP groups. Primary diagnosis was identified using ICD-9 and ICD-10 codes. Dependent variables included cost per patient (categorized by laboratory, radiology, and total cost), hospital bed days, and disposition status. All data were analyzed across both the study and control groups.

All data available in the MHS M2 database were deidentified. Consequently, the analysis was exempt from the federal regulations for the protection of human research participants, and an institutional review board approval was not necessary. The data were obtained from the M2 database after completing the data user agreement with the MHS.

Results

The results of the 3-month resident survey showed that the most utilized features and appreciated features on the RHAP were the ones that gave residents instant access to national and institutional guidelines. Implementation of the RHAP resulted in residents' self-reporting statistically significant improvement in accessibility of these guidelines (p < 0.05) and departmental contact information (p = 0.0039).

Objective data collected, using the M2 database, showed that during the 26 months of the study, a total of 912 patients were admitted to the Neurology service. There were a total of 430 admissions pre-RHAP, an average of 33 admissions per month, with a range of 22–44 admissions per month. Post-RHAP there were 482 admissions, an average of 37 admissions per month, with a range of 26–50 admissions per month. The average age of the patients admitted pre-RHAP was 53 years and post-RHAP was 51 years. The gender of patients admitted pre-RHAP was 211 male and 219 female and post-RHAP was 224 male and 258 female. There were a total of 1,578 patient bed days with a total cost of $8,112,432.96 pre-RHAP and 1,592 patient bed days and $7,858,971.57 in total costs post-RHAP.

Patient bed days, total cost per patient, and cost per patient bed day all decreased post-RHAP. During the same time period, there were no changes or increases in these areas with the control group (Table 1). Multivariate linear regression analysis was used to control for the most common neurological diagnoses and patient demographics. There was a statistically significant decrease in costs per patient for Laboratory tests (−$102.34/patient) (p < 0.02). Other statistically significant factors affecting the cost of laboratory were patient bed days and the primary diagnosis of demyelinating disease (Table 2). There was also a strong trend toward decreased costs of radiology studies per patient (−$78.30/patient) (p = 0.07) post-RHAP. There were other independent variables that had demonstrated statistical significance in regard to radiology costs, including primary diagnosis (demyelinating disease, stroke, mental illness, headache, and myasthenia gravis), patient age, and total hospital bed days (Table 3). Our analysis found that patient disposition status was not affected by use of the RHAP.

Table 1.

Neurology and Internal Medicine Comparison Before and After RHAP Implementation

	ADMISSIONS	TOTAL BED DAYS	TOTAL COST	AVERAGE BED DAYS PER PATIENT	AVERAGE COST PER PATIENT	AVERAGE COST PER BED DAY
Neurology
Before	430	1,578	8,112,432.96	3.67	18,866.12	5,140.96
After	482	1,592	7,858,971.57	3.30	16,304.92	4,936.54
% Difference	10.79	0.88	−3.23	−11.11	−15.71	−4.14
Internal medicine
Before	3,427	16,934	72,334,965.11	4.94	21,107.37	4,271.58
After	3,542	20,290	86,650,690.94	5.73	24,463.77	4,270.61
% Difference	3.25	16.54	16.52	13.74	13.72	−0.02

RHAP, Resident Handbook Application.

Table 2.

Full Cost of Laboratory: Multivariable Analysis Using Linear Regression

	CHANGE	LOWER CI	UPPER CI	p
Full cost laboratory
RHAP (before vs. after)	−102.34	−185.02	−19.67	0.02
Gender	4.09	−79.48	87.67	0.92
Age	0.41	−1.74	2.57	0.70
Stroke	63.77	−62.51	190.06	0.32
Mental illness	−20.42	−330.01	289.16	0.89
Brain infection	232.19	−71.61	536.01	0.13
Headache	−43.48	−217.48	130.50	0.62
Demyelinating diseases	274.47	83.54	465.40	0.0
Epilepsy	13.09	−152.32	178.51	0.87
Myasthenia	164.66	−149.17	478.50	0.30
Bed days	−155.20	−171.07	−139.32	0.0

Change = Increase or Decrease in U.S. dollars if one positive integer is added. A binary system was assigned to RHAP (0 = Before, 1 = After implementation), Gender (0 = Male, 1 = Female), and Diagnosis (0 = No Diagnosis, 1 = Diagnosis); actual Age and Bed Days were used. Diagnosis identified by ICD-9, 10 codes. Ex. Patients with a Diagnosis of Demyelinating Disease had an increase in the cost of laboratories by $274.47 when all other variables were held constant.

CI, confidence interval; RHAP, Resident Handbook Application.

Table 3.

Cost of Radiology: Multivariable Analysis Using Linear Regression

	CHANGE	LOWER CI	UPPER CI	p
Full cost laboratory
RHAP (before vs. after)	−78.31	−164.44	7.81	0.07
Gender	0.37	−86.70	87.45	0.99
Age	2.03	−0.26	4.27	0.07
Stroke	1,310.43	1,178.86	1,442.08	0.00
Mental illness	−605.96	−928.50	−283.43	0.00
Brain infection	35.58	−280.93	352.10	0.82
Headache	321.07	139.80	502.33	0.00
Demyelinating diseases	1,524.84	1,325.92	1,723.75	0.00
Epilepsy	−128.85	−301.18	43.48	0.14
Myasthenia	396.66	69.70	723.62	0.02
Bed days	−45.05	−61.59	−28.51	0.00

Change = Increase or Decrease in US Dollars if one positive integer is added. A binary system was assigned to RHAP (0 = Before, 1 = After implementation), Gender (0 = Male, 1 = Female), and Diagnosis (0 = No Diagnosis, 1 = Diagnosis); actual Age and Bed Days were used. Diagnosis identified by ICD-9, 10 codes. Ex. Patients with a Diagnosis of Demyelinating Disease had an increase in the cost of radiology by $1,524.84 when all other variables were held constant.

Discussion

The RHAP had a significant and positive effect upon multiple measures of performance. Total cost per patient decreased by $2,561 (15.7%) with implementation of the RHAP. This was a savings of $1,234,402 for the 482 patients admitted post-RHAP (Table 1). Controlling for the decreased number of bed days post-RHAP the savings/patient was $474.42 (4.14%). Multivariate linear regression demonstrated a statistically significant decrease in the cost of laboratory studies ($102/patient decrease) (p = 0.02) and a strong trend in decreased costs for radiology studies ($78/patient decrease) (p = 0.07). This was a savings of $180 per patient with a total savings of $86,760 on the 482 patients admitted post-RHAP implementation. The use of the RHAP drove down costs from a variety of perspectives without affecting patient disposition status. Patient demographics, primary diagnosis, and bed days (Tables 2 and 3) were all controlled.

Diagnoses for the pre- and post-RHAP groups were tracked to control of differences in diagnoses being a cause for the cost differentials. A significant decrease in high cost diagnoses after implementation of the RHAP could explain the outcomes observed. As shown in Tables 2 and 3 “high cost diagnoses” such as demyelinating disease, myasthenia gravis, and brain infection were identified. These diagnoses were controlled using multivariable analysis. After controlling for diagnosis there continued to be significant differences in costs between the pre- and post-RHAP groups. The author concludes that differences in diagnoses, such as more “high cost diagnoses” pre-RHAP implementation, were not the cause of the cost differences between the pre- and post-RHAP groups.

Many subspecialties were considered as a control group, IM was chosen for several reasons. Subspecialty areas such as Cardiology are more prone to extraneous factors such as department directives that could change practice. This, in turn, could cause a decrease or increase in spending, hospital bed days, etc. without a change in a hospital wide trend. Using all admissions to the Medical Center was also considered; however, this would include surgical specialties that were considered a poor comparison group to neurology. This group would also include the study group. IM is a large and heterogeneous inpatient admitting service. For this reason, it was the best way to look for hospital wide trends that cross specialties and diagnoses that could have possibly explained the differences between the pre- and post-RHAP groups, such as command directives or hospital initiatives. As shown in Table 1 no such trends were seen. IM actually exhibited a cost increase in several categories. This suggests that the cost differences seen post-RHAP were not part of a hospital wide trend.

Our 3-month survey results showed a statistically significant improvement in residents' ability to access the institutional and national neurological guidelines (p < 0.05) and departmental contact information (p = 0.0039). It is believed that this access helps account for the statistically significant overall decrease in spending with laboratory tests and the significant downward trend in radiology costs post-RHAP. It is well documented that residents typically increase the cost of hospitalizations,^9

–12 most notable in the areas of diagnostic tests especially laboratory and radiology testing.^13

–16 This application appeared to reduce unnecessary laboratory and radiology tests ordered by residents, tests that otherwise may have been ordered in this medical training environment. RHAP decreased hospital spending and more importantly saved patients from unnecessary medical tests and diagnostic procedures.

A decrease in average (mean) hospital bed days from 3.66 to 3.30 per patient (9.8% reduction) (Table 1) was another positive outcome of the RHAP implementation. Although this reduction did not rise to the level of statistical significance, it is worth noting. The use of the RHAP may have expedited patient workups and disposition processes by decreased unnecessary laboratory and radiology tests that may have extended the patients stay.

An extensive PubMed search was conducted on six occasions through the course of the study. Although mobile applications have been used with subjective studies where they are used by patients to get better outcomes and in Graduate Medical Education,^{5,17

–22} no studies were found similar to the current study. This is the first study to objectively demonstrate statistically significant data showing a decrease in hospital spending with use of a mobile application by physicians.

Our study has several limitations. First, the M2 database is an administrative database. Any such database may have coding errors embedded within its datasets. Second, this is a retrospective observational study that is subject of normal biases, including selection bias. Third, multiple variables, including changes in resident physicians and staff physicians pre- and post-RHAP, were not controlled.

Conclusions

Development and implementation of the RHAP were essentially at no cost and only involved the time of a few hard working residents. The Application's design and implementation can be easily copied and implemented in almost any medical setting. The correlations shown above holds promise to eliminate unnecessary laboratory and imaging studies in the medical training environment; therefore, decreasing spending and freeing up stressed hospital recourses. This study should prompt consideration for using a similar application in a variety of medical and practice settings. There are many mobile applications that contain medical guidelines and diagnostic tools. However, this is the first study to demonstrate that a mobile application resulted in statistically significant cost savings and reported improvements in patient care.

Footnotes

Acknowledgments

The author gratefully acknowledges the support of the Neurology Department at Walter Reed Bethesda with special thanks to Drs. Brett Theeler and Philip Eye, without which the present study could not have been completed.

Disclosure Statement

No competing financial interests exist.

References

Fingas

. Two-thirds of Americans now have smartphones. Available at www.engadget.com/2014/02/11/two-thirds-of-americans-now-have-smartphones (last accessed December 10, 2015 ).

Perez

. comScore: In U.S. mobile market, Samsung, android top the charts; apps overtake web browsing. Available at https://techcrunch.com/2012/07/02/comscore-in-u-s-mobile-market-samsung-android-top-the-charts-apps-overtake-web-browsing (last accessed December 10, 2015 ).

WhatTech.. 1.5 million apps are live on the google play store. Available at www.whatech.com/mobile-apps/news/30223-1-5-million-apps-are-live-on-the-google-play-store (last accessed December 15, 2015 ).

Mashable.. App store stats: 400 million accounts, 650,000 apps. Available at http://mashable.com/2012/06/11/wwdc-2012-app-store-stats/#o9Kx6zE6qkq0 (last accessed December 10, 2015 ).

Franko

, Tirrell

. Smartphone app use among medical providers in ACGME training programs. J Med Syst, 2012; 36:3135–3139.

Sharpe

3rd , Kendrick

, Strickland

, Dodd

3rd . The radiology resident iPad toolbox: An educational and clinical tool for radiology residents. J Am Coll Radiol, 2013; 10:527–532.

Gibson

, Nguyen

, Davidson

, Panagiotopoulos

. Effectiveness of an educational handbook in improving psychiatry resident knowledge of second-generation antipsychotics. Acad Psychiatry, 2015; 39:154–159.

Shaw

, Tan

. Integration of mobile technology in educational materials improves participation: Creation of a novel smartphone application for resident education. J Surg Educ, 2015; 72:670–673.

Charlson

, Karnik

, Wong

, McCulloch

, Hollenberg

. Does experience matter? A comparison of the practice of attendings and residents. J Gen Intern Med, 2005; 20:497–503.

10.

Khaliq

, Huang

, Ganti

, Invie

, Smego

Jr . Comparison of resource utilization and clinical outcomes between teaching and nonteaching medical services. J Hosp Med, 2007; 2:150–157.

11.

Shine

, Beg

, Jaeger

, Pencak

, Panush

. Association of resident coverage with cost, length of stay, and profitability at a community hospital. J Gen Intern Med, 2001; 16:1–8.

12.

Bekmezian

, Chung

, Yazdani

. Staff-only pediatric hospitalist care of patients with medically complex subspecialty conditions in a major teaching hospital. Arch Pediatr Adolesc Med, 2008; 162:975–980.

13.

Garber

, Fuchs

, Silverman

. Case mix, costs, and outcomes. Differences between faculty and community services in a university hospital. N Engl J Med , 1984; 310:1231–1237.

14.

Busby

, Leming

, Olson

. Unidentified educational costs in a university teaching hospital: An initial study. J Med Educ, 1972; 47:243–253.

15.

Martz

, Ptakowski

. Educational costs to hospitalized patients. J Med Educ, 1978; 53:383–386.

16.

Schroeder

, O'Leary

. Differences in laboratory use and length of stay between university and community hospitals. J Med Educ, 1977; 52:418–420.

17.

Boulos

, Wheeler

, Tavares

, Jones

. How smartphones are changing the face of mobile and participatory healthcare: An overview, with example from eCAALYX. Biomed Eng Online, 2011; 10:24.

18.

Wallace

, Clark

, White

. “It's on my iPhone”: Attitudes to the use of mobile computing devices in medical education, a mixed-methods study. BMJ Open, 2012; 2:e001099.

19.

Chu

, Erlendson

, Sun

, Alva

, Clemenson

. Mobile computing in medical education: Opportunities and challenges. Curr Opin Anaesthesiol, 2012; 25:699–718.

20.

Smeds

, Thrush

, Mizell

, Berry

, Bentley

. Mobile spaced education for surgery rotation improves National Board of Medical Examiners scores. J Surg Res, 2016; 201:99–104.

21.

Woolford

, Clark

, Strecher

, Resnicow

. Tailored mobile phone text messages as an adjunct to obesity treatment for adolescents. J Telemed Telecare, 2010; 16:458–461.

22.

Hale

, Capra

, Bauer

. A Framework to Assist health professionals in recommending high-quality apps for supporting chronic disease self management: Illustrative assessment of type 2 diabetes apps. JMIR Mhealth Uhealth, 2015; 3:e87.