Market analysis and Geographic Information Systems (GIS) in transplantation

Market analysis and Geographic Information Systems (GIS)

in transplantation

Joel Thomas Adler

Disclosures

• Wilmar Chocolates are hand made, hand cut, hand wrapped, and very tasty

• Appleton, WI• Please enjoy

http://bit.ly/1uWEsCh for chocolate ID

Kidney and liver transplantation

• 620,000 people with ESRD in the US

• 16,000 waiting for liver transplantation

• Scarcity and allocation

• Liver transplant rates greatly vary across country

Organ Procurement and Transplant Network

Deceased donor liver transplant rates per 100 patient years on the waiting list

Market competition influences renal transplantation and outcomes

Joel T. Adler, MD,1,2 Rosh K. V. Sethi, BS,3 Heidi Yeh, MD,2,3 James F. Markmann, MD, PhD,2,3

and Louis L. Nguyen, MD, MBA, MPH1,3,4

1Center for Surgery and Public Health at Brigham and Women’s Hospital2Division of Transplant Surgery, Massachusetts General Hospital

3Harvard Medical School4Division of Vascular & Endovascular Surgery, Brigham and Women’s Hospital

Competition varies by Donor Service Area (DSA)

Outcomes worse in DSAs of higher competition for deceased donors

Patient mortality Graft failure

HR (95% CI) P value HR (95% CI) P value

Competition

All patients0.99 (0.92, 1.07) 0.78 1.07 (0.99, 1.15) 0.08

Living donor0.94 (0.80, 1.11) 0.48 0.99 (0.85, 1.15) 0.89

Deceased donor1.11 (1.02, 1.21) 0.01 1.18 (1.09, 1.28) <0.0001

• Likely not a center-specific effect• Absolute differences small• Better outcomes than dialysis

“Markets” and scarce resources

• Donor service areas (DSAs) functioning as an individual “market”

• Increasing market competition associated with riskier organs and worse survival, but better than alternative

• How can we use geography to better understand and optimize?

Adler et al Ann Surg 2014Halldorson et al Liver Trans 2013

Markets and GIS: why does this matter?

• Allocation linked to geography

• Provide insight into utilization patterns

• Justify our definition of DSA markets

• Larger discussions of allocation policy

Gentry AJT 2013

Geographic Information Systems (GIS)

• Integrating geographic information

• Long history to understand problems in healthcare

• Strength in data layering, combinations, interpolation, and spatial associations

GIS in HSR

Surgery• Estimating burden of

disease in LMIC (Tollefson TT Laryngoscope 2014)

• Gunshot trauma (Livingston DH J Trauma Acute Care Surg 2014)

• Variation in care (Vassileva C J Heart Valve Dis 2012)

• Technology adoption (Sethi J Vasc Surg 2013)

Everybody else• Access to stroke care (Adeoye

O Stroke 2014)• Need and access in CKD

(Rodriguez RA J Nephrol 2013)• Environmental exposures in

children (Harrison F Int J Health Geogr 2014)

• Health disparities and mammography utilization (Ayanian JZ JNCI 2013)

Market competition and density in liver transplantation: relationship

to volume and outcomeJoel T. Adler, MD1,2, Heidi Yeh, MD2,3,

James F. Markmann, MD, PhD2,3, and Louis L. Nguyen, MD, MBA, MPH1,3,4

1Center for Surgery and Public Health at Brigham and Women’s Hospital2Division of Transplant Surgery, Massachusetts General Hospital

3Harvard Medical School4Division of Vascular & Endovascular Surgery, Brigham and Women’s Hospital

Market competition and density in transplantation

• Transplant centers unevenly distributed in the DSAs

• Competition and transplant center density are likely important

• Incorporate the spatial arrangements into models to better understand access and outcomes

ACS 2014, NESS 2014

Density and Organization: Average Nearest Neighbor (ANN)

• Geocoded transplant centers• Categorized as clustered, random, or dispersed;

single as a special case• Considers spatial arrangement more than distance

Average Nearest Neighbor by DSA (Liver)

NESS 2014

Market characteristicsAbsolute Nearest Neighbor (ANN)

Variable All DSAs(n = 446)

Single(n = 150)

Clustered(n = 164)

Random (n = 93)

Dispersed(n = 39)

P value

Population (millions)

5.28 (3.59 – 8.72)

3.33(2.46 – 4.71)

6.25(4.73 – 10.8)

6.34(4.49 – 9.35)

8.70(7.37 – 17.0)

<0.0001

Liver transplant centers

2 (1 – 3)

1(1 - 1)

2(2 – 3)

2(2 – 4)

3(2 – 4)

<0.0001

HHI 0.56 (0.50 – 1.00)

1.00(1.00 – 1.00)

0.52(0.39 – 0.61)

0.50(0.42 – 0.53)

0.51 (0.31 – 0.59)

<0.0001

New listings 166 (98 – 299)

72(40 – 117)

220 (161.5 – 425.5)

232(155 – 401)

285(168 – 602)

<0.0001

Deceased organ donors

139 (88 – 217)

87(52 – 131)

169.5(108 – 266)

151(114 – 257)

211(171 – 444)

<0.0001

Liver transplants 87.5 (55 – 162)

48(23 – 73)

109(74.5 – 205)

131(85 – 166)

179(88 – 436)

<0.0001

MELD score at transplant

25.1 ± 0.1 23.3 ± 0.2 26.6 ± 0.2 25.8 ± 0.3 23.9 ± 0.3 <0.0001

LDRI Unadjusted 1.51

(1.44 – 1.57)1.44

(1.36 – 1.53)1.52

(1.45 – 1.60)1.52

(1.49 – 1.59)1.54

(1.51 – 1.61)<0.0001

Adjusted 1.37 (1.31 – 1.43)

1.32(1.25 – 1.39)

1.39(1.32 – 1.44)

1.40(1.36 – 1.44)

1.39(1.35 – 1.43)

<0.0001

NESS 2014

Liver transplants performed

Variable IRR (95% CI) P valueAdult liver transplant centers 1.03 (1.01 – 1.06) 0.04Competition (inverse HHI) 1.33 (1.03 – 1.69) 0.03New listings (100s) 1.14 (1.10 – 1.17) <0.0001Donors (100s) 1.25 (1.17 – 1.32) <0.0001Population (millions) 1.04 (1.00 – 1.07) 0.02Geography (by ANN)

Single Ref -Clustered 1.25 (1.13 – 1.38) <0.0001Random 1.24 (1.09 – 1.41) 0.001

Dispersed 1.43 (1.10 – 1.85) 0.007MELD score (at transplant) 0.97 (0.96 – 0.98) <0.0001Adjusted LDRI 3.35 (2.54 – 4.43) <0.0001

NESS 2014

Patient and graft outcomes

MortalityVariable HR (95% CI) P value

Liver transplant centers 1.01 (0.98 - 1.04) 0.68Competition (inverse HHI) 0.99 (0.77 – 1.29) 0.96New listings (100s) 1.02 (0.99 – 1.04) 0.16Donors (100s) 1.05 (0.99 – 1.10) 0.04Population (millions) 0.99 (0.98 – 1.01) 0.08Geography (by ANN)

Single Ref -Clustered 1.02 (0.91 – 1.14)Random 1.03 (0.91 – 1.17) 0.65

Dispersed 1.03 (0.91 – 1.17) 0.62Adjusted LDRI 1.56 (1.47 – 1.66) <0.0001

Graft failureHR (95% CI) P value

1.05 (1.01 – 1.08) 0.012.17 (1.64 – 2.86) <0.00010.94 (0.91 – 0.97) <0.00011.13 (1.07 – 1.19) <0.00011.03 (1.01 – 1.05) 0.0002

Ref -1.51 (1.34 – 1.71) <0.00011.31 (1.14 – 1.51) 0.00021.01 (0.87 – 1.17) 0.901.68 (1.56 – 1.80) <0.0001

NESS 2014

Conclusions

• Market variables and ANN are most important for graft survival

• Transplant center density has a measurable impact on liver transplants and patient and graft survival

• Increasing the number of liver transplant centers within a DSA could provide better access to liver transplantation

Market and socioeconomic factors in the conduct of kidney transplantation

Joel T. Adler, MD1,2, Heidi Yeh, MD2,3, James F. Markmann, MD, PhD2,3, and Louis L. Nguyen, MD, MBA, MPH1,3,4

1Center for Surgery and Public Health at Brigham and Women’s Hospital2Division of Transplant Surgery, Massachusetts General Hospital

3Harvard Medical School4Division of Vascular & Endovascular Surgery, Brigham and Women’s Hospital

Market and socioeconomic factors in the conduct of kidney transplantation

• Kidney transplants dependent on market factors (SRTR)

• Socioeconomic factors affect access to kidney transplantation (US ACS)

• These factors may be spatially correlated to better understand kidney transplantation

ASC 2015 (submitted)

Competition by ZCTA

Competition in the United States

ASC 2015 (submitted)

Spatial regression

• Classically linear (housing prices in Manhattan)• Spatial error– Omitted (spatially correlated) covariate– Errors are not independent

• Spatial lag– “Diffusion” process: events in one place predict and

increased likelihood of events in other areas– Observations and errors are not independent

• Dependent on weights (queen, rook, K nearest neighbor…)

Kidney transplants and SES factors are spatially related

----------------------------------------------------------------------- Variable Coefficient Std. Error z-value Probability ----------------------------------------------------------------------- CONSTANT 40.56982 4.683726 8.661868 0.0000000 HHI_HSA_IN 27.54952 1.712949 16.0831 0.0000000 CROWDED 1.16356 0.2482719 4.686636 0.0000028 POVERTY -0.1164395 0.1354502 -0.8596483 0.3899829 LOW_EDUCAT -0.05833486 0.1134684 -0.5141065 0.6071775 HIGH_EDUCA 0.3939563 0.1205663 3.267548 0.0010850 UNEMPLOYME 0.8934559 0.2050763 4.356701 0.0000132 MPV -3.4443e-005 7.431e-006 -4.633904 0.0000036 MHI -9.5672e-005 7.181e-005 -1.332215 0.1827895 LAMBDA 0.459516 0.02209419 20.79805 0.0000000-----------------------------------------------------------------------

ASC 2015

Conclusions

• Competition and SES effects diffuse among neighboring HSAs

• Spatial autocorrelation plays a role in factors influencing kidney transplantation

• Consider these issues in planning transplant center location and organ sharing

ASC 2015 (submitted)

Conclusions: this does matter!

• Allocation and utilization are linked to geography

• Utilization patterns and cost

• Justify our definition of DSA markets

• Allocation policy

Gentry AJT 2013

Resources

• Center for Geographic Analysis (http://www.gis.harvard.edu/)

• Open GeoDA (https://geodacenter.asu.edu/ogeoda)

• ESRI ArcGIS (http://www.esri.com/software/arcgis)

Three boys? Why not?

| ￣￣￣￣￣￣ | | JANUARY 24 | | | | (SIGN BUNNY) | | ＿＿＿＿＿__| (\__/) || (• ㅅ• ) || / 　 づ

Projects

• ZIP codes and SES of donors and recipients• Spatial organization of centers– Kidney transplants– Liver transplants

• Competition maps and access to transplantation• Market competition density index• Provider-induced demand• Disparities in donation rates (Bode)

GIS tools

• Data display and interpretation• Combining data and interpolating• Hotspot/outlier analysis• Organization of points• Spatial regression

Low-quality kidneys are used in more competitive DSAs

Variable OR (95% CI) P value

Competition None 1.00 -

Low 1.20 (1.08, 1.32) 0.0005

Medium 1.05 (0.95, 1.16) 0.33

High 1.39 (1.26, 1.52) <0.0001

Average Nearest Neighbor by DSA (Kidney)

ACS 2014

Hotspot/outlier analysis for competition: Local Indicator of Spatial Autocorrelation (LISA)

ASC 2015 (submitted)

Display and interpretation

Overview

• Transplantation and markets: competition and outcomes

• Geographic Information Systems (GIS)• GIS in HSR• GIS techniques and how we’ve used them• Future directions