PreprintPDF Available

Analysis of Surgical Wait Times in Nova Scotia

Authors:
Carlo Carandang at Canada Revenue Agency
Carlo Carandang
  • Canada Revenue Agency
Gregory Horne at Atlantic Data Analytics 7AD
Gregory Horne
  • Atlantic Data Analytics 7AD
William Wells at Saint Mary's University
William Wells
C. Stokes at Nova Scotia Community College
C. Stokes
Preprints and early-stage research may not have been peer reviewed yet.
Download file PDF
Read file
Download citation

Abstract

OBJECTIVES: Surgical wait times in Nova Scotia are at crisis levels, as patients are waiting too long on the wait list. This can lead to worsening illness course, and in some cases premature death. This project started with the objective of addressing the surgical waitlist crisis, as people in Nova Scotia are waiting months to years before receiving their life- saving surgeries. METHODS: This study involved obtaining a clinical dataset from an open data portal on the internet, loading the data into our analytical environment, exploring the data using analytical software, cleaning the data, determining feature variables (independent variables), and utilizing the data to train models for prediction of the outcome variable(s). R was utilized to perform exploratory, descriptive, and predictive statistical analysis on the surgical wait times dataset in Nova Scotia, for the time period 2014 to 2016. RESULTS: A bivariate multiple linear regression model was constructed with two dependent variables (consult_90th and surgery_90th), representing the 90th percentiles for each instance of a surgical specialty’s wait time, added together to give the combined surgical wait time and one independent variable. CONCLUSION: Although we can give an estimate of average wait times in Nova Scotia based on specialty from the model, only the waittimes from the following specialties are statistically significant: General Surgery, Dental, Opthalmology, Orthopaedic, and Otolaryngology (ENT) at 95% confidence. However, as the model is statistically significant with all the coefficients, then it can be assumed that the model can predict with 95% confidence the wait times from any specialty, even if the individual coefficient is not statistically significant.
Content uploaded by Carlo Carandang
Author content
All content in this area was uploaded by Carlo Carandang on Mar 18, 2018
Content may be subject to copyright.
Analysis of Surgical Wait Times in Nova Scotia
Carlo&Carandang
1
,&Gregory&Horne
2
,&William&Wells1,&Catherine&Stokes2
Corresponding&Author:&Carlo&Carandang,&MD&
Email:&carandangc@gmail.com&
ORCID:&0000-0002-2008-9642&
This project was supported by Saint Mary’s University, MSc Program in Computing & Data
Analytics and NSCC Institute of Technology
The statistical analysis and code for this study is located at this repository:
https://github.com/Carlo-Carandang/Nova_Scotia_Surgical_Wait_Times
&
1
Saint Mary’s University, MSc Program in Computing & Data Analytics, 923 Robie Street, Halifax, Nova Scotia,
Canada, B3H 3C3
2
NSCC, Institute of Technology Campus, 5685 Leeds Street, Halifax, Nova Scotia B3K 2T3
Analysis of Surgical Wait Times in Nova Scotia
Abstract
OBJECTIVES:!Surgical!wait!times!in!Nova!Scotia!are!at!crisis!levels,!as!patients!are!waiting!
too!long!on!the!wait!list.!This!can!lead!to!worsening!illness!course,!and!in!some!cases!
premature!death.!This!project!started!with!the!objective!of!addressing!the!surgical!waitlist!
crisis,!as!people!in!Nova!Scotia!are!waiting!months!to!years!before!receiving!their!life-
saving!surgeries.!!
METHODS:!This!study!involved!obtaining!a!clinical!dataset!from!an!open!data!portal!on!the!
internet,!loading!the!data!into!our!analytical!environment,!exploring!the!data!using!
analytical!software,!cleaning!the!data,!determining!feature!variables!(independent!
variables),!and!utilizing!the!data!to!train!models!for!prediction!of!the!outcome!variable(s).!
R!was!utilized!to!perform!exploratory,!descriptive,!and!predictive!statistical!analysis!on!the!
surgical!wait!times!dataset!in!Nova!Scotia,!for!the!time!period!2014!to!2016.!!
RESULTS:!A!bivariate!multiple!linear!regression!model!was!constructed!with!two!
dependent!variables!(consult_90th!and!surgery_90th),!representing!the!90th!percentiles!
for!each!instance!of!a!surgical!specialty’s!wait!time,!added!together!to!give!the!combined!
surgical!wait!time!and!one!independent!variable.!!
CONCLUSION:!Although!we!can!give!an!estimate!of!average!wait!times!in!Nova!Scotia!
based!on!specialty!from!the!model,!only!the!waittimes!from!the!following!specialties!are!
statistically!significant:!General!Surgery,!Dental,!Opthalmology,!Orthopaedic,!and!
Otolaryngology!(ENT)!at!95%!confidence.!However,!as!the!model!is!statistically!significant!
with!all!the!coefficients,!then!it!can!be!assumed!that!the!model!can!predict!with!95%!
confidence!the!wait!times!from!any!specialty,!even!if!the!individual!coefficient!is!not!
statistically!significant.!
Surgical!wait!times!in!Nova!Scotia!are!at!crisis!levels,!as!patients!are!waiting!too!long!on!the!
wait!list.!This!can!lead!to!worsening!illness!course,!and!in!some!cases!premature!death.!
This!project!started!with!the!objective!of!addressing!the!surgical!waitlist!crisis,!as!people!in!
Nova!Scotia!are!waiting!months!to!years!before!receiving!their!life-saving!surgeries.!The!
Nova!Scotia!government!website!only!gives!numbers!about!wait!times!that!are!difficult!to!
interpret.!This!research!team!wanted!to!look!for!correlations!of!the!variables!in!the!surgical!
wait!times!dataset.!In!fact,!every!province!has!a!wait!times!dataset!that!is!updated!every!3!
months,!and!post!the!data!to!the!Canadian!Institute!for!Health!Information!public-facing!
website.!The!data!goes!back!to!the!beginning!of!the!recordings,!which!is!roughly!around!the!
early!2010s.!
The!data!set!for!surgical!wait!times!in!Nova!Scotia!has!a!dozen!columns!and!approximately!
7000!rows!when!the!dataset!was!obtained!for!the!years!2014!to!2016.!It’s!basically!an!
Excel!spreadsheet,!and!is!not!in!any!relational!database!tables.!It!maps!surgical!wait!times!
based!on!date,!hospital,!Health!Region!Zone,!specialty,!procedure,!and!doctor.!
Research!shows!that!surgical!wait!times!are!associated!with!significant!morbidity!and!
mortality!risk1–6.!Morbidity!risk!is!the!probability!of!the!illness!worsening,!while!mortality!
risk!is!the!probability!of!dying!from!the!illness.!It!is!widely!known!that!the!surgical!wait!
times!in!Nova!Scotia!are!at!crisis-levels!currently.!Patients!in!Nova!Scotia!are!waiting!too!
long!on!average!to!receive!their!procedure.!However,!it!is!not!known!how!the!data!can!
inform!the!improvement!of!services,!and!hence!morbidity!and!mortality.!Therefore,!it!is!
important!to!analyze!the!wait!times!data!using!advanced!statistical!and!data!analytics!to!
help!provide!more!insights!to!help!decrease!morbidity!and!mortality!from!waiting!for!
surgical!procedures.!
Although!the!most!urgent!surgeries!get!scheduled!first,!waiting!too!long!for!surgery!can!be!
detrimental!to!one’s!health!because!waiting!too!long!is!a!burden!on!the!patient!and!their!
family,!waiting!too!long!can!lead!to!death,!and!waiting!too!long!can!lead!to!worsening!of!the!
illness.!Indeed,!while!the!most!urgent!surgeries!get!scheduled!first,!waiting!too!long!for!
surgery!can!be!detrimental!to!one’s!health.!
Ultimately,!the!researchers!wanted!to!find!the!factors!associated!with!surgical!wait!times,!
so!that!new!insights!can!be!obtained!to!decrease!the!wait!times,!and!hence!save!lives!and!
prevent!illness!deterioration.!
Methods
This!study!involved!obtaining!a!clinical!dataset!from!an!open!data!portal!on!the!internet,!
loading!the!data!into!our!analytical!environment,!exploring!the!data!using!analytical!
software,!cleaning!the!data,!determining!feature!variables!(independent!variables),!and!
utilizing!the!data!to!train!models!for!prediction!of!the!outcome!variable(s).!
The!research!team!divided!the!work!into!the!following!tasks:!
• Project!topic!selection!(C.C.,!G.H.,!W.W.,!C.S.)!
• Dataset!selection!(C.C.,!W.W.)!
• Data!cleaning!(C.C.,!G.H.)!
• Exploratory!analysis!and!multiple!linear!regression!in!R!(G.H.,!W.W.)!
• Multiple!linear!regression!and!polynomial!regression!in!Python!(C.C.!and!G.H.)!
• Visualization!of!data!and!graphs!(G.H.)!
• Research,!report!writing,!and!editing!(C.C.,!G.H.,!W.W.,!C.S.)!
Initially,!the!data!had!to!be!cleaned,!as!there!were!many!missing!values.!This!was!
performed!manually,!as!the!column!and!rows!were!populated!haphazardly,!and!therefore!
had!to!inspect!and!note!the!missing!values,!row!by!row!(12!columns!and!about!7000!rows).!
Subsequently,!business!and!clinical!rules!were!followed!to!fill!in!missing!values,!utilizing!
the!clinical!domain!expertise!of!one!of!the!authors!(C.C.)!who!had!practiced!as!a!physician!
in!Nova!Scotia!for!several!years.!As!an!example!of!missing!values,!each!of!the!surgical!
procedures!had!to!be!classified!into!the!surgical!specialty!to!which!it!belonged,!as!many!of!
the!rows!were!missing!this!information.!In!addition,!there!were!missing!values!for!hospital,!
Health!Region!Zone,!procedure!performed!by!surgeon,!and!specialty!of!the!surgeon.!Any!
missing!values!not!obtainable!within!the!dataset!were!scraped!from!the!internet!to!
populate!the!missing!values,!such!as!hospital!information!and!information!on!the!surgeon.!
Once!the!dataset!was!cleaned!and!missing!values!were!replaced!with!inferred!values!from!
domain!expertise!and/or!web!scraping,!exploratory!statistical!analysis!was!performed!on!
the!surgical!wait!times!dataset!in!Nova!Scotia.!The!research!team!was!trying!to!determine!if!
there!were!any!correlations!between!the!numerous!independent!variables!(i.e.!specialty,!
period,!facility,!etc.)!and!the!dependent!variable!(surgical!wait!times).!This!exploratory!
analysis!was!the!beginning!of!an!ongoing,!in-depth!analysis!of!the!Nova!Scotia!Surgical!Wait!
Times,!to!help!to!improve!patient!care!through!data!analysis!and!statistical!modelling.!
R!was!utilized!to!perform!exploratory,!descriptive,!and!predictive!statistical!analysis!on!the!
surgical!wait!times!dataset!in!Nova!Scotia,!for!the!time!period!2014!to!2016.!
Analytical Environment
Software
Data!analyses,!visualizations,!and!production!of!this!report!were!undertaken!using!
software!with!the!specifications!detailed!below.!!
• Anaconda!v0.0.0!
• Jupyter!Notebook!v0.0.0!
• pdfTeX!(TeXlive!2015)!v3.14159265-2.6-1.40.16!
• Pandoc!v1.16.0.2!
• R!v3.4.4!
• R!Studio!v1.1.383!
• R!Markdown!v2!
Hardware
Data!analyses,!visualizations,!and!production!of!this!report!were!undertaken!using!
hardware!with!the!specifications!detailed!below:!!
Apple6MacBook6Air!Processor:!1.6!GHz!Intel!Core!i5;!Number!of!Processors:!1;!Total!
Number!of!Cores:!2;!L2!Cache!(per!Core):!256!KB;!L3!Cache:!3!MB;!Memory:!8!GB!1600!
MHz!DDR3;!Operating!System:!Apple!Mac!OS!X!v10.0!
Apple6Macbook6Pro!Processor:!3.1!GHz!Intel!Core!i7;!Number!of!Processors:!1;!Total!
Number!of!Cores:!2;!L2!Cache!(per!Core):!256!KB;!L3!Cache:!3!MB;!Memory:!16!GB!1857!
MHz!DDR3;!Graphics:!Intel!Iris!Graphics!6100!1536!MB;!Operating!System:!Apple!Mac!OS!X!
v10.13!
Hewlett6Packard6Spectre613!Processor:!2.5GHz!Intel!Core!i7;!Number!of!Processors:!1;!
Total!Number!of!Cores:!4;!Memory:!8!GB!DDR3;!Graphics:!Intel!HD!Graphics!520;!Operating!
System:!Xubuntu!Linux!v16.04!LTS!
Acquire the Surgical Wait Times Data
The!surgical!wait!times!data!was!obtained!from!the!Nova!Scotia!Government!Open!Data!
Portal.!Here!are!the!steps!to!acquire!the!dataset:!
1) Navigate!to!the!Open!Data!Portal!
2) Select!the!Data!Catalogue!
3) Search!‘surgical!wait!times’!
4) Select!Surgical!Wait!Times!
5) Select!the!Export!option!
6) Select!CSV!
7) Save!the!data!file!to!the!project!subdirectory!
(nova_scotia_wait_times/data_analysis/data).!
8) Load!the!data!from!a!tab-separated-values!formatted!file!with!column!headers!
Wrangle the Data
Once!the!data!was!loaded,!the!feature!names!of!the!data!set!were!determined:!
[1]!Period![2]!Specialty![3]!Procedure![4]!Provider!
[5]!Zone![6]!Facility![7]!Year![8]!Quarter!
[9]!Consult_Median![10]!Consult_90th![11]!Surgery_Median![12]!Surgery_90th!
The!dataset!contained!6843!observations!and!12!features.!Viewing!a!sample!of!the!dataset,!
it!was!determined!that!multiple!variables!were!categorical,!and!the!wait!times!were!the!
only!continuous!variables.!Here!are!the!definitions!of!the!12!features!in!the!dataset:!
Feature!
Definition!
Period!
Time!period!
Specialty!
Surgical!specialty!
Procedure!
Surgical!procedure!
Provider!
Surgeon!
Zone!
Healthcare!Zone,!1!to!4!
Facility!
Hospital!
Year!
Year!
Quarter!
Quarter!(3-months)!
Consult_Median!
Maximum!time!that!50%!of!patients!recently!waited!for!consult!
Consult_90th!
Maximum!time!that!90%!of!patients!recently!waited!for!consult!
Surgery_Median!
Maximum!time!that!50%!of!patients!recently!waited!for!surgery!
Surgery_90th!
Maximum!time!that!90%!of!patients!recently!waited!for!surgery!
The!surgical!specialty!types!were!parsed!as!follows:!specialty,!all!specialties,!cardiac,!
dental,!general,!neurosurgery,!obstetrics/gynaecology,!ophthalmology,!oral!and!
maxillofacial,!oral!maxillofacial,!orthopaedic,!otolaryngology!(ent),!plastic,!thoracic,!
urology,!vascular.!In!total,!there!are!15!surgical!specialty!types!and!159!surgical!procedure!
types.!
Next!was!the!scatterplots!of!the!features!paired!against!one!another:!
Figure.!Features!as!pairs!in!scatterplots.!
!
The!exploratory!analysis!gave!us!an!initial!signal!that!specialty!may!be!correlated!with!wait!
times,!as!the!scatterplots!of!the!various!combinations!of!the!variables!showed!that!the!
specialties!showed!a!‘banded’!pattern!when!it!was!paired!and!plotted!versus!wait!times,!
and!some!specialties!appeared!to!have!greater!variation!in!wait!times!than!others.!The!
‘banded’!pattern!occurred!due!to!the!specialty!variable!being!categorical!(discrete!
datapoints).!Because!of!this!banded!pattern,!specialty!was!chosen!as!the!independent!
variable,!and!consult!and!surgery!times!were!chosen!as!the!dependent!variable.!Therefore,!
after!this!exploration!of!the!dataset!for!features!to!study,!only!observations!with!surgical!
procedure!type!‘all’!were!extracted:!
feature!
missing_count!
nonmissing_count!
consult_90th!
12!
284!
consult_median!
12!
284!
facility!
296!
0!
period!
0!
296!
procedure!
0!
296!
provider!
0!
296!
quarter!
296!
0!
specialty!
0!
296!
surgery_90th!
0!
296!
surgery_median!
0!
296!
year!
296!
0!
zone!
296!
0!
The!minimum,!maximum,!average,!standard!deviation,!and!total!combined!wait!days!by!
consultation!and!surgery!specialty.!Average!wait!days!is!the!median!number!of!days!on!the!
surgical!wait!list:!
minimum!
maximum!
average!
sigma!
total!
observations!
66!
198!
157!
49!
702!
5!
148!
1032!
327!
319!
7006!
16!
65!
2234!
177!
298!
14432!
56!
155!
949!
252!
236!
3081!
10!
64!
882!
199!
149!
9573!
41!
115!
2875!
392!
497!
16779!
33!
171!
620!
421!
159!
4332!
11!
162!
1365!
662!
318!
26539!
38!
136!
1081!
390!
258!
11910!
25!
151!
738!
372!
186!
5598!
15!
73!
449!
179!
134!
1307!
6!
61!
819!
219!
170!
6002!
22!
112!
685!
307!
242!
2151!
6!
Rows!with!missing!values!were!removed!before!training!the!prediction!model.!
Visualise the Data
After!wrangling!the!data,!the!data!was!visualized:!
Figure!1.!
!
Figure!1!shows!the!total!wait!time!in!person!years,!grouped!by!surgical!specialty.!Cardiac!
surgery!has!the!lowest!value,!while!orthopaedic!has!the!highest.!
Figure!2.!
!
Figure!2!shows!the!median!total!wait!time!in!days,!grouped!by!surgical!specialty.!Cardiac!
surgery!has!the!lowest!value,!while!orthopaedic!has!the!highest.!
Figure!3.!
!
Figure!3!shows!the!wait!time!distribution!in!days,!grouped!by!surgical!specialty.!Notice!
cardiac!surgery!and!thoracic!surgery!at!the!lower!end,!and!orthopaedic!at!the!higher!end.!
Figure!4.!
!
Figure!4!shows!the!histogram!of!the!frequencies!of!the!wait!times.!Most!people!are!waiting!
between!100!to!200!days!for!their!surgical!procedure.!
Figure!5.!
!
Figure!5!shows!the!histogram!of!the!frequencies!of!the!wait!times,!grouped!by!consultation!
wait!times!and!surgery!wait!times.!Most!people!are!waiting!between!50!days!and!200!days!
for!a!consulation,!and!between!50!to!250!days!for!a!surgical!procedure.!
It!was!determined!from!the!dataset!that!the!total!wait!time!to!receive!a!surgical!procedure!
is!the!sum!of!the!consultation!wait!time!and!the!surgical!wait!time.!The!domain!expert!who!
had!worked!in!the!system!(C.C.)!for!several!years!interpreted!the!total!wait!time!for!a!
surgical!procedure!as!the!sum!of!the!following!wait!times:!
*!Wait!time!to!see!a!family!doctor,!who!will!determine!if!a!surgery!consultation!is!needed!
(family!doctor!wait!time)!*!Wait!time!to!consult!with!a!surgeon!(consult!wait!time)!*!Wait!
time!for!the!surgical!procedure!(surgery!wait!time)!
So!the!total!wait!time!to!receive!a!surgical!procedure!is!as!follows:!
Total Wait Time = Family Doctor Wait Time + Consult Wait Time + Surgery Wait Time
But!this!dataset!is!missing!the!Family!Doctor!Wait!Time,!so!the!estimated!wait!times!for!
surgical!procedures!in!this!analysis!is!missing!the!Family!Doctor!Wait!Time.!Therefore,!any!
estimated!prediction!of!wait!times!from!this!study!is!an!underestimate!of!the!real!wait!
time.!
Results
Build the Model
Prior!to!building!the!statistical!model!the!baseline!factor!was!set!to!‘general!surgery’!
instead!of!the!default!‘cardiac!surgery’!to!determine!the!impact,!if!any,!on!the!linear!
regression!model!with!regards!to!the!null!hypothesis.!
A!bivariate!multiple!linear!regression!model!was!constructed!with!two!dependent!
variables!(consult_90th!and!surgery_90th),!representing!the!90th!percentiles!for!each!
instance!of!a!surgical!specialty’s!wait!time,!added!together!to!give!the!combined!surgical!
wait!time!and!one!independent!variable.!
A summary of the multiple linear regression model is as follows:
Call:!lm(formula!=!specialty90!~!specialty)!
Residuals:!
Min!
1Q!
Median!
3Q!
Max!
-536.39!
-144.57!
-66.16!
70.76!
2366.55!
Coefficients:!
Estimate!Std.!
Error!
t!value!
Pr(>!
(Intercept)!
257.71!
38.80!
6.643!
1.68e-10!***!
specialtycardiac!
-117.31!
135.51!
-0.866!
0.387417!
specialtydental!
180.16!
82.30!
2.189!
0.029447!*!
specialtyneurosurgery!
50.39!
99.67!
0.506!
0.613607!
specialtyobstetrics/gynaecology!
-24.23!
59.68!
-0.406!
0.685083!
specialtyophthalmology!
250.74!
63.71!
3.935!
0.000106!***!
specialtyoral!maxillofacial!
136.10!
95.75!
1.421!
0.156338!
specialtyorthopaedic!
440.68!
61.02!
7.222!
5.19e-12!***!
specialtyotolaryngology!(ent)!
218.69!
69.83!
3.131!
0.001930!**!
specialtyplastic!
115.49!
84.41!
1.368!
0.172388!
specialtythoracic!
-39.88!
124.72!
-0.320!
0.749385!
specialtyurology!
15.10!
73.05!
0.207!
0.836358!
specialtyvascular!
100.79!
124.72!
0.808!
0.419727!
Signif.!codes:!0!|!***!0.001!|!**!0.01!|!*!0.05!|!‘.’!0.1!|!‘’!1!
Residual!standard!error:!290.3!on!271!degrees!of!freedom!
Multiple!R-squared:!0.2383,!Adjusted!R-squared:!0.2046!
F-statistic:!7.066!on!12!and!271!DF,!p-value:!3.522e-11!
From the above summary of the model, the following statistical measures were extracted:
degrees!of!freedom:!13!and!271!
p-value!of!the!model:!3.522271910^{-11}!
residual!standard!error:!290.3315348!
F-statistic:!7.0658736!
F-critical:!1.7583858!
An analysis of variance (ANOVA) of the linear regression model was used to validate the
model:
Analysis!of!Variance!Table!
Response:!specialty90!
Measure!
Df!
Sum!Sq!
Mean!Sq!
F!value!
Pr(>F)!
specialty!
12!
7147193!
595599!
7.0659!
3.522e-11!***!
Residuals!
271!
22843240!
84292!
!
!
Interpretation
The!residual!standard!error!(standard!error!of!the!estimate)!is!290.33,!where!it!is!the!
standard!deviation!of!the!variation!of!observations!around!the!regression!line!(it!is!the!
standard!deviation!of!the!regression!model).!So!waitimes!average!can!be!estimated!as!+-
2(290.33)!=!580.66.!As!this!value!is!large!when!compared!to!the!average!waittimes!in!the!
sample,!then!the!variation!of!observed!y!values!from!the!regression!line!is!also!large.!For!
future!research,!we!should!look!for!other!variables!which!can!explain!more!of!the!variation!
in!waittimes!(ie.!costs!for!procedures,!facilities!funding,!staffing!levels,!etc.).!
R^2!is!0.2383,!but!we!do!have!to!utilize!the!adjusted!R^2,!as!we!have!multiple!independent!
variables.!The!adjusted!R^2!is!0.2046,!where!20.46%!of!the!variation!in!waittimes!is!
explained!by!the!variation!in!specialty,!taking!into!account!the!sample!size!and!number!of!
independent!variables.!
Next,!we!want!to!determine!if!the!model!is!significant.!Hypothesis!testing!is!set!up!as!
follows:!
Hypotheses: H0: ( 1 ) = ( 2 ) = … = ( k ) = 0 (no linear relationship)
HA: at least one ( i ) ≠ 0 (at least one independent variable affects y)
F statistic = (SSR/k)/(SSE/(n-k-1))
Since!F!=!7.07!is!in!the!rejection!region!(it!is!greater!than!the!F(critical)!=!1.76),!we!reject!
the!null!hypothesis!at!alpha!=!0.05,!and!accept!the!alternative!hypothesis!that!at!least!one!
independent!variable!affects!total!surgical!wait!times!at!95%!confidence.!According!to!the!
model!the!H0!(null!hypothesis)!should!be!rejected!in!favour!of!HA!(alternative!hypothesis).!
We!determine!that!the!model!is!significant.!
In!addition,!we!conclude!that!the!regression!model!does!explain!a!significant!portion!of!the!
variation!in!waittimes.!This!conclusion!is!also!confirmed!by!the!F-statistic!with!a!p-value!=!
3.522271910^{-11},!and!therefore!it!is!less!than!the!alpha!value!of!0.05,!and!we!can!also!
conclude!from!this!p-value!that!the!model!does!explain!a!significant!portion!of!the!variation!
in!waittimes.!
An!analysis!of!variance!(ANOVA)!of!the!linear!regression!model!was!used!to!validate!the!
model.!For!example,!the!degrees!of!freedom,!sum!of!squares,!and!mean!squared!can!easily!
be!retrieved!for!both!the!speciality!and!residuals!parameters,!and!the!F-statistic!for!the!
specialty!parameter.!The!values!of!the!F-statistic!(7.0658736)!and!F-critical!(1.7880147)!
indicate!the!H0!(null!hypothesis)!should!be!rejected!in!favour!of!HA!(alternative!
hypothesis).!
Next,!we!extract!the!coefficients!so!the!linear!regression!model!equation!can!be!
constructed.!Here!is!the!prediction!model:!
waittimes = 257.71 - 117.31(Cardiac Surgery) + 180.16(Dental) + 50.39(Neurosurgery) -
24.23(Obstetrics/Gynaecology) + 250.74(Opthalmology) + 136.10(Oral Maxillofacial) +
440.68(Orthopaedic) + 218.69(Otolaryngology (ENT)) + 115.49(Plastic Surgery) -
39.88(Thoracic Surgery) + 15.10(Urology) + 100.79(Vascular Surgery)
General!Surgery!is!the!default,!where!all!the!other!specialty!variables!are!0.!The!model!
predicts!the!following!for!average!wait!times!by!specialty:!
Although!we!can!give!an!estimate!of!average!wait!times!in!Nova!Scotia!based!on!specialty!
from!the!model,!only!the!waittimes!from!the!following!specialties!are!statistically!
significant:!General!Surgery,!Dental,!Opthalmology,!Orthopaedic,!and!Otolaryngology!(ENT)!
at!95%!confidence.!However,!as!the!model!is!statistically!significant!with!all!the!
coefficients,!then!it!can!be!assumed!that!the!model!can!predict!with!95%!confidence!the!
wait!times!from!any!specialty,!even!if!the!individual!coefficient!is!not!statistically!
significant.!
Conclusion
In!this!project,!we!started!with!the!analysis!for!specialty!vs!wait!times,!as!the!exploratory!
analysis!gave!us!an!initial!signal!that!specialty!may!be!correlated!with!wait!times,!as!the!
scatterplots!of!the!various!combinations!of!the!variables!showed!that!the!specialties!
showed!a!‘banded’!pattern!when!it!was!paired!and!plotted!versus!wait!times,!and!some!
specialties!appeared!to!have!greater!variation!in!wait!times!than!others.!The!‘banded’!
pattern!occurred!due!to!the!specialty!variable!being!categorical!(discrete!datapoints).!
We!used!multiple!linear!regression!for!specialty!vs.!wait!times,!and!found!that!our!model!
explained!20%!of!the!variation!in!the!dependent!variable!(wait!times).!Although!we!can!
give!an!estimate!of!average!wait!times!in!Nova!Scotia!based!on!specialty!from!the!model,!
only!the!wait!times!from!the!following!specialties!are!statistically!significant:!General!
Surgery,!Dental,!Opthalmology,!Orthopaedic,!and!Otolaryngology!(ENT)!at!95%!confidence.!
However,!as!the!model!is!statistically!significant!with!all!the!coefficients,!then!it!can!be!
assumed!that!the!model!can!predict!with!95%!confidence!the!wait!times!from!any!
specialty,!even!if!the!individual!coefficient!is!not!statistically!significant.!
In!summary,!our!model!for!specialty!vs.!total!wait!times!(consult!plus!surgery!wait!times)!
was!statistically!significant,!as!indicated!by!our!F-statistic!with!a!p!<!0.05.!This!F-statistic!
analysis!was!performed!with!all!of!the!specialties!included!in!the!model,!without!backward!
elimination.!
!
!
References
1.!Coughlin!SMP,!Guidolin!K,!Fortin!D,!Frechette!E,!Malthaner!R,!Inculet!R.!Is!it!safe!to!wait?!
The!effeft!of!surgical!wait!time!on!survival!in!patients!with!non-small!lung!cell!cancer.!Can!J!
Surg.!2015;58(6):414–18.!!
2.!Malaisrie!SC!KJ!McDonald!E.!Mortality!while!waiting!for!aortic!valve!replacement.!Ann!
Thorac!Surg.!2014;98(5):1564–70.!!
3.!Kulkarni!GS!AP!Urbach!DR.!Longer!wait!times!increase!overall!mortality!in!patients!with!
bladder!cancer.!J!Urol.!2009;182(4):1318–24.!!
4.!Rexius!H!OA!Brandrup-Wognsen!G.!Mortality!on!the!waiting!list!for!coronary!artery!
bypass!grafting:!Incidence!and!risk!factors.!Ann!Thorac!Surg.!2004;77(3):769–74.!!
5.!Koomen!EM!KJ!Hutten!BA.!Morbidity!and!mortality!in!patients!waiting!for!coronary!
artery!bypass!surgery.!Eur!J!Cardiothorac!Surg.!2001;19(3):260–5.!!
6.!Everhart!JE!DK!Lombardero!M.!Increased!waiting!time!for!liver!transplantation!results!in!
higher!mortality.!Transplantation.!1997;64(9):1300–6.!!
ResearchGate has not been able to resolve any citations for this publication.
Is it safe to wait? The effect of surgical wait time on survival in patients with non-small cell lung cancer
Article
Full-text available
  • Nov 2015
Background: The effect of surgical wait times on survival in patients with non-small cell lung cancer (NSCLC) remains largely unknown. Our objective was to determine the effect of surgical wait time on survival and incidence of upstaging in patients with stage I and II NSCLC. Methods: All patients with clinical stage I and II NSCLC who underwent surgical resection in a single centre between January 2010 and December 2011 were reviewed. Analysis was stratified based on preoperative clinical stage. We assessed the effect of wait time on survival using a Cox proportional hazard model with wait time in months as a categorical variable. Incidence of upstaging at least 1 stage was assessed using logistic regression. Results: We identified 222 patients: 180 were stage I and 42 were stage II. For stage I, wait times up to 4 months had no significant effect on survival or incidence of upstaging. For stage II, patients waiting between 2 and 3 months had significantly decreased survival (hazard ratio 3.6, p = 0.036) and increased incidence of upstaging (odds ratio 2.0, p = 0.020) than those waiting 0 to 1 month. For those waiting between 1 and 2 months, there was no significant difference in survival or upstaging. Conclusion: We did not identify an effect of wait time up to 4 months on survival or upstaging for patients with stage I NSCLC. For patients with stage II disease, wait times greater than 2 months adversely affected survival and upstaging.
View
Mortality While Waiting for Aortic Valve Replacement
Article
  • Sep 2014
  • ANN THORAC SURG
Background: Severe symptomatic aortic stenosis (AS) is associated with high mortality without intervention. The impact of waiting time for aortic valve replacement (AVR), either surgically or transcatheter, has not been reported. Methods: From January 2008 to December 2012, we identified 1,005 patients with severe symptomatic AS. AVR was recommended for 823 patients (82%). Of these 823 patients, 721 (87.6%) underwent AVR. We modeled overall survival (OS) since AVR recommendation or intervention date using Cox and multistate models. Results: Overall, the median (first, third quartiles) waiting time until operation was 2.9 (1.3, 5.1) weeks. Mortality at these times was lower (p<0.001) in the AVR group (1.2%, 0.3%, 1.7%, respectively) than in the group that did not receive AVR (6.9%, 2.9%, 9.8%, respectively). Thirty-day mortality after AVR was 3.9% (3.2% surgical AVR [SAVR] and 7.0% transcatheter AVR [TAVR]). In patients receiving AVR, waiting time was not associated with increased mortality. Mortality while waiting for AVR was 3.7% and 11.6% at 1 and 6 months, respectively. Mortality while waiting for TAVR was higher than that for SAVR (1-, 6-, and 12-month mortality of 3.7%, 8.0%, and 9.6%, respectively, in SAVR group and 3.8%, 23.3%, and 27.5%, respectively, in TAVR group; p<0.001). Conclusions: Some patients do not receive AVR in a timely fashion, and prolonged waiting time for AVR is associated with mortality greater than the AVR operative mortality. Although waiting time was not associated with poor operative outcomes after AVR, many patients may die while waiting for AVR. Patients should receive AVR on a semiurgent, not elective, basis.
View
Increased waiting time for liver transplantation results in higher mortality
Article
  • Nov 1997
  • TRANSPLANTATION
Waiting time to liver transplantation (LTx) has dramatically lengthened, but the proportion of candidates who die awaiting transplantation has not increased. We evaluated whether longer waiting time for LTx candidates increases mortality. A cohort of candidates listed for LTx between 1990 and 1993 by three large transplantation programs was followed for 2 years. The exposure measure was ABO blood type, which is not inherently related to outcome, but is a major determinant of waiting time. The main outcome measure was 2-year mortality, as evaluated by logistic regression analysis that controlled for differences in clinical status at the time of evaluation for LTx. The 308 candidates with type O blood waited longer for LTx (median 109 days) than the 399 candidates with other blood types (median 58 days) (P=0.001). Candidates listed for LTx with type O blood had better clinical status at evaluation, but then had higher pretransplantation mortality (13.3%) than other candidates (7.0%) (P=0.005). Blood group O candidates had higher 2-year mortality (26.6%) than other candidates (22.1%), which on multivariate analysis resulted in a mortality odds ratio at 2 years of 1.52 (95% confidence interval=1.04-2.23). With the difference in median waiting time between blood groups increasing from 44 days in the first year to 108 days in the third year, the 2-year mortality odds ratio also rose from 0.94 to 1.97. When compared with LTx candidates with other blood types, blood type O candidates have longer waiting times and higher pretransplantation mortality, which results in higher 2-year mortality.
View
Longer Wait Times Increase Overall Mortality in Patients With Bladder Cancer
Article
  • Sep 2009
  • J UROLOGY
We used population level data to determine the impact of extended wait times on the survival of patients who underwent radical cystectomy for bladder cancer. We identified 2,535 patients who underwent cystectomy for bladder cancer in Ontario, Canada between 1992 and 2004 using administrative databases. A Cox proportional hazards model accounting for patient, pathological and health services variables that could affect wait times was created to assess the impact of wait time on survival. The tumor stage specific impact of waiting for cystectomy was also assessed. Cox regression analysis that modeled wait time using cubic splines was used to determine a maximum wait time within which optimal care can be provided. Median wait time from transurethral bladder resection to cystectomy was 50 days. Unadjusted and adjusted analyses demonstrated that prolonged wait times were significantly associated with a lower overall survival rate. The relative hazard of death with increasing wait times appeared greater for low stage vs high stage cancers. The cubic splines regression analysis revealed that the risk of death began to increase after 40 days. Treatment delay between transurethral bladder tumor resection and radical cystectomy resulted in worse overall survival. The effect of wait time was greatest in lower stage lesions. The suggested maximum wait time from transurethral bladder tumor resection to cystectomy was 40 days. Further studies assessing disease-free survival are required to corroborate these findings.
View
Morbidity and mortality in patients waiting for coronary artery bypass surgery
Article
  • Apr 2001
To describe morbidity and mortality in patients waiting for coronary artery bypass graft (CABG) surgery and to assess determinants for the occurrence of these complications. A prospective cohort study was carried out in a tertiary referral general teaching hospital. Three hundred and sixty consecutive patients with a priority of routine or urgent who were accepted for CABG or CABG with additional valve surgery were evaluated. Follow-up began from the moment of acceptance until the procedure took place for cardiac death, myocardial infarction and unstable angina requiring hospital admission. The median (25-75th percentile) waiting time in the two priority groups was 100 (79-119) days for the routine group and 69 (38-91) days for the urgent group. Overall, eight patients died, seven suffered a myocardial infarction, and 33 episodes of unstable angina requiring immediate hospitalization occurred. The majority of events took place during the first 30 days on the waiting list. Unstable angina less than 3 months before acceptance was identified as an independent predictor (hazard ratio 2.5, 95% confidence interval 1.2-5.1) for complications during the wait. The prognostic value of smoking and familial cardiovascular disease was found to vary depending on the priority assigned to the patient. Complications occur relatively early during the time on the waiting list. If complications in coronary heart disease cannot be predicted more accurately, the only way to diminish the complication rate is drastic reduction of waiting times.
View
Mortality on the waiting list for coronary artery bypass grafting: Incidence and risk factors
Article
  • Mar 2004
  • ANN THORAC SURG
Insufficient capacity for coronary artery bypass grafting results in waiting times before operation, prioritization of patients and, ultimately, death on the waiting list. We aimed to calculate waiting list mortality and to identify risk factors for death on the waiting list. The study included 5,864 consecutive patients accepted for elective coronary artery bypass grafting (78% male; mean age, 66 +/- 9 years). The patients were categorized at acceptance into three priority groups: imperative (39%), urgent (36%), or routine (25%). Waiting list mortality was calculated and compared between groups, and risk factors were identified by Poisson regression. Median waiting time for the whole population was 55 days. Seventy-seven patients (1.3%) died, corresponding to a mortality rate of 5.8 deaths per 100 patient-years. The mortality rate per 100 patient-years was highest for those in the imperative group, 15.1 deaths, compared with 5.3 deaths in the urgent group and 3.2 in the routine group (p < 0.001). Independent risk factors were male sex (p = 0.032), Cleveland Clinic risk score (p = 0.005), impaired left ventricular ejection fraction (p = 0.007), unstable angina pectoris (p = 0.001), concomitant aortic valve disease (p = 0.002), priority group (p < 0.001), and time after acceptance (p = 0.019). The mortality risk increased with time after acceptance by 11% a month. Long waiting lists for coronary artery bypass grafting are associated with considerable mortality. The risk of death increases significantly with waiting time. Sex, unstable angina, perioperative risk, impaired left ventricular function, and concomitant aortic valve disease are independent risk factors and should be considered at triage.
View