The impact of temperature on mortality in Ludhiana … impact of temperature on mortality in Ludhiana City, India: A time series analysis Suresh Kumar Sharma et al International Journal

INTERNATIONAL JOURNAL OF ENVIRONMENTAL SCIENCES Volume 3, No 6, 2013

© Copyright by the authors - Licensee IPA- Under Creative Commons license 3.0

Research article ISSN 0976 – 4402

Received on January 2013 Published on August 2013 2323

The impact of temperature on mortality in Ludhiana City, India: A time

series analysis Suresh Kumar Sharma1, Rajesh Kumar2, Rashmi Aggarwal1, PVM Lakshmi2, Kanchan Jain1

1- Department of Statistics, Panjab University, Chandigarh, India

2- School of Public Health, Post Graduate Institute of Medical Education and Research,

Chandiagrh, India

[email protected]

doi:10.6088/ijes.2013030600048

ABSTRACT

The widely used generalized additive model (GAM) is a flexible and effective technique for

conducting non-linear regression analysis in time-series studies. There is strong evidence that

episodes of extremely hot or cold temperatures are associated with increased mortality in

many parts of the world. Time series designs are extremely useful to examine association

between daily apparent temperature and daily mortality counts. Hence, the effect of

temperature on mortality was studied in Ludhiana city of Punjab in Northern India. As a part

of the Health Effects Institute (HEI), Boston USA, project, Meteorological and mortality data

was obtained for the years 2002-2004. Sahnewal Airport in Ludhiana City records

temperature, dew point, wind speed and relative humidity at 8.30 AM, 11.30 AM, and 5.30

PM. Daily death records were obtained from the civil registration system in Ludhiana. The

association between temperature and mortality was established using the generalized additive

model (GAM) with penalized spline smoothers at 8,4,4 degrees of freedom (df) in R software

with mortality count (excluding accidents) as a dependent variable. Smoothers for day of the

week, relative humidity and wind speed were included in the model. With rapid

industrialization and extreme temperature variations in Ludhiana city, the temperature was

significantly associated with mortality. Sensitivity analysis shows that elderly (>65 years)

population was much affected with temperature variations, particularly, during winter season.

The study shows there is need to improve the overall registration system of deaths. Cause

specific analysis was not possible as cause of death was not clearly mentioned in most of the

deaths.

Keywords: Generalized additive model, relative humidity, windspeed, mortality, sensitivity

analysis, time series analysis.

1. Introduction

For many years, the effect of temperature on mortality has been the subject of numerous

studies, mostly examining the impact of extreme weather events (Hastie and Tibshirani,1990

and Ha et al.,2011). Numerous studies have shown a positive association between daily

mortality and temperature during extreme hot and cold weather. Early studies found a relation

between peak of daily mortality and a concomitantly peak of temperatures (Carson et al.,2006

and Kim et al.,2006). Mortality due to extreme temperatures have also been reported by many

authors (Braga et al.,2001). In these studies, the elderly and those suffering from underlying

medical conditions such as circulatory and respiratory diseases were found to be more

vulnerable (Hajat et al,2002 and Goldberg et al.,2011). Most of the existing research has been

conducted in Europe (McKee,1989, Baker,2008 and Ballester et al.,2011), Spain (Ballester et

The impact of temperature on mortality in Ludhiana City, India: A time series analysis

Suresh Kumar Sharma et al International Journal of Environmental Sciences Volume 3 No.6 2013

2324

al.,1997), Dutch (Huynen,2001), Germany (Hoffmann,2008), Canada (Goldberg et al.,2011)

and America (Ellis and Nelson,1978, Curriero,2002, Anderson and Bell,2009, Ostro et

al.,2009 and Basu and Malig,2011). Very few studies in Asia, South Korea (McKee,1989 and

Ha et al.,2011), India (Kumar et al.,2010) which is rapidly developing and where effects of

temperature on mortality are increasing have been conducted. The effect of various

temperature indicators on different mortality categories in a subtropical city of Brisbane have

also been conducted recently (Yu et al.,2011). Reviews of epidemiologic studies from 2001

to 2008 have been conducted by Basu (Basu,2006). Modelling temperature effects on

mortality using multiple segmented relationships with common break points was also

conducted (Muggeo,2008).

Time Series analysis using Generalized Additive Model (GAM) (Hastie and Tibshirani,1990)

show an association between temperature and mortality across a range of less extreme

temperatures after smoothing the effects of day of the week, relative humidity and wind

speed. In this paper, the authors describes the temperature-mortality association for year

2002-2004 in Ludhiana City of northern India by estimating the relative risks of mortality

using GAM with quasi-poisson function for mortality due to natural causes (excluding

accidents) as the dependent variable and temperature as the independent variable with

penalized spline smoothers for day of the week, relative humidity and wind speed. Current

and recent day’s temperatures were the weather components most strongly predictive of

mortality, and mortality risk generally decreased as temperate increased from the coldest days

to a certain threshold temperature, above which mortality risk increased as temperature

increased to highest level or decreased to the lowest level. The model developed in this

analysis is potentially useful for projecting the consequences of climate change scenarios and

offering insights into susceptibility to the adverse effects of temperature.

2. Material and methods

The present study was carried out in Ludhiana city, which is the largest city of Punjab in

northern India. The Indian meteorological Department, New Delhi, records meteorological

data at the Sahnewal airport of Ludhiana city. Temperature, relative humidity, dew point,

wind speed and wind direction was recorded daily at 8.30 AM, 11.30 AM and 5.30 PM.

Meteorological data was available for all the 365 days in a year. Municipal corporation

records mortality data at two Centre’s which register deaths for zone A and B, C and D of the

city respectively. Government and Private hospitals report deaths that occur in their premises

and heads of families reports deaths which happen at home or places other than hospital to

the municipal corporation office in their respective zone. The data available in the registers

by registration number, name of the deceased, father’s name, age, sex, date of death, cause of

death, place of death and address of the deceased were entered in computer for analysis.

Standard operating procedures were developed for maintaining the quality of data. During

data entry, 2% of the entries were randomly compared with the records, each week in each of

the two zones. At the end of data entry, a 100% recheck was again carried out. It has been

found out that sex and age were not mentioned in 207(0.7%) and 21(0.1%) of the deaths.

2.1 Statistical analysis

The data was examined for quality and consistency. Summary statistics, mean, standard

deviation, minimum, maximum and range were computed as per Table1 for temperature,

relative humidity and wind speed. Overall in the 3 years period 2002-2004, 28007 deaths

were registered with an average of 25.4 deaths per day. The age-sex distribution of the deaths



2325

is shown in Table 2. There were more deaths among males (65%) and in the above 45 years

age group (67%). Only 787(2.8%) of the deaths were due to accidents and these were

excluded from analysis. Most of the deaths (71.2%) fall under the category of “symptoms,

signs and abnormalities classified elsewhere” due to improper certification of the cause of

death. Table 1: Summary Statistics of meteorological parameters

Sta

tist

ics Temperature(0C) Relative Humidity (%) Wind Speed (km/hour)

2002 2003 2004 Overall 2002 2003 2004 Overall 2002 2003 2004 Overa

ll

Mea

n

25.87 24.99 25.91 25.59 56.85 59.86 57.66 58.12 7.50 6.41 6.70 6.87

S.D

.

7.72 8.37 7.65 7.93 16.78 21.58 19.16 19.30 5.22 4.83 4.50 4.88

Min

imu

m

8.13 6.20 8.37 6.20 11.00 10.33 13.67 10.33 .00 .00 .00 .00

Ma

xim

um

41.53 41.47 40.60 41.53 93.67 99.67 97.67 99.67 42.67 35.67 27.33 42.67

Table 2: Age and sex distribution of registered deaths

Age

Group

YEAR

Total 2002 2003 2004

Male Female Male Female Male Female

<1year 380

149

375

120

318

143 1485(5.3%)

1-4 yrs 73

39 63

34 52

37 298(1.1%)

5-44 yrs 1680

739 1620

742 1668

715 7164(25.8%)

45-64 yrs 1313

815 1897

963 1990

931 8409(30.3%)

65+ yrs 1889

1408 2102

1486 2065

1473 10423(37.5%)

Total 5835

3150

6057

3345

6093

3299 27779*(100%) *age and sex missing in 207 and 21 records respectively (27779+207+21=28007)

In Northern India, Ludhiana city is the largest City of Punjab state with population of 1.5

million (projected from census 2001) and having area of 135 square km. This city is located

at a distance of about 100 km from Chandigarh and about 360 km from New Delhi (Figure



2326

2). The industrial city of Ludhiana has high temperature variations from 00C to 45-480C with

a mean temperature of around 280C (Figure 1). To describe the daily variation in weather data

smoother plots for relative humidity and wind speed were drawn for years 2002-2004 with 20

df (Figure 3 and Figure 4). It is observed that smooth plot of relative humidity followed

similar trends for almost three years; however, wind speed variations were observed quite

frequently during three years. Extreme values of wind speed varies from 15 km/h to 38 km//h

in all the three years. These variations could not be controlled even when df were increased

up to 30. In this part of the region, some dust storms are always expected, particular, during

summer (June-July) and winter (December-January) season.

Figure 1: Box Plot for Max, Min and Mean Temperature

Figure 2: Map of Ludhiana City



2327

3. Development of GAM model

First of all, base core model was developed in R package (Wood,2006 and Kumar et

al. ,2010). The Generalized cross validation (GCV) score for penalized spline at 8,4,4 df was

less than that for penalized spline of 6,3,3 df and natural spline model with 8,4,4 df as well as

for 6,3,3 df. Thus, we selected penalized spline with 8,4,4 df . Sensitivity analysis was carried

out separately for deaths in the 65+ year’s age group and also for gender discrimination.

Smaller number of deaths in 0-4 year’s age group (1.1%) did not allow us to perform

sensitivity analysis for this age group.

0 200 400 600 800 1000

20

40

60

80

100

Days

Rela

tive H

um

idity(%

)

RH 8:30RH 11:30RH 17:30AVG.RH

0 200 400 600 800 1000

05

10

15

20

25

30

35

Days

Win

s S

peed(K

m/h

r)

WS 8:30WS 11:30WS 17:30AVG.WS

Figure 3 and 4 : Smooth Plot of Relative Humidity and Smooth Plot of Wind Speed

3.1 Partial Autocorrelation Function Plots (PACF)

A characteristic feature of time series is that the observations are ordered through time; a

consequence of this is autocorrelation, that is, an underlying pattern between observations

from one time period to the next within a time series (Barron,1992, Peng and Dominici,2008



2328

and Ha et al., 2011). We can define autocorrelation at log k as the correlation between

observations k time periods apart. This can be estimated for lags k = 0, 1, 2 . . . as

where xt denotes the observation at time t and denotes the sample mean of the series .

The partial autocorrelation at lag k is defined as the partial correlation between observations k

time periods apart. That is, it is the correlation between xt and xt-k after taking out the effect

of the intervening observations xt-1, . . ., xt-k+1. It can be estimated using a fast computational

algorithm. The collection of autocorrelations at lags k = 1, 2, . . . is known as the

autocorrelation function (ACF). Similarly, the collection of partial autocorrelations is known

as the partial autocorrelation function (PACF).

3.2 Residual Plot

A residual plot is a graph that shows the residuals on the vertical axis and the independent

variable on the horizontal axis. If the points in a residual plot are randomly dispersed around

the horizontal axis, a linear regression model is appropriate for the data; otherwise, a non-

linear model is more appropriate. The difference between the observed value of the

dependent variable (y) and the predicted value (ŷ) is called the residual (e). Each data point

has one residual. Thus, Residual = Observed value - Predicted value. A plot of residuals

versus predicted response is essentially used to spot possible heteroscedasticity (non-constant

variance across the range of the predicted values), as well as influential observations

(possible outliers). Usually, we expect such plot to exhibit no particular pattern (a funnel-like

plot would indicate that variance increase with mean). Plotting residuals against one predictor

can be used to check the linearity assumption.

Since a GAM is just a penalized GLM, residual plots are checked, exactly as for a GLM. The

distribution of scaled residuals was examined, marginally, and plotted against covariates and

fitted values. In R residuals (model) extracts residuals and gam.check (model) produces

simple residual plots, and summary of convergence information.Plot (model, residuals =

TRUE) plots smooth terms with partial residuals overlaid. In the sequel, we have worked out

the PACF and Residual plots for Lag0, Lag1, Lag2 and Lag3 (Schwartz, 2000) along with

predicted plot of mortality in R using, PS(8,4,4) model.

3.2.1 Model: Effect of Temperature on Mortality: GeneralizedAdditive Model (GAM)

Log (E(mortality))=Temperature +s(day effect)+ s(relative humidity)+ s(wind speed)

1. Generalized Additive Model (GAM) with penalized spline smoothers in R.

2. Quasi-Poisson function with mortality from all natural causes as the dependent

variables

3. Smoothers for day effect, relative humidity and wind speed

4. Day of week terms (from Monday through Sunday).

5. Exposure at single-day lags of 0 to 3 days



2329

Table 3: Effect of Temperature on Mortality in Ludhiana 2002-2004 (all deaths)

Mean temp 95% CI

Lag

Effects N β-Coeff.

Relative

Risk P-value Lower Upper

Lag 0 1096 0.015443 1.166992586 .0000118* 1.08945 1.25005

Lag 1 1095 0.011836 1.125649272 .0000421* 1.06391 1.19097

Lag 2 1094 0.007806 1.081187528 0.0048 * 1.02421 1.14133

Lag 3 1093 0.005579 1.057375612 0.0406 ** 1.00244 1.11532

* (p < .01), **(p < .05)

-20

2

Day

Re

sid

ua

l

Jan02 Jul02 Jan03 Jul03 Jan04 Jul04 Dec04

0 5 10 15 20 25 30

-0.1

00

.00

0.1

0

Lag

Pa

rtia

l A

CF

10

20

30

40

50

Day

Num

be

r o

f ep

isod

es


Figure 5: Residual and PACF plots for PS (8,4,4) model along with predicted plot of

mortality (Lag0)

It is evident from PACF and Residual plots that there is significant effect of temperature on

mortality after smoothing the effects of day, relative humidity and wind speed. The predicted



2330

plot of mortality for Lag0 can be used to determine episodes of mortality during a particular

day (Figure 5). Except few outliers, particularly, during the months of January and July

(2002-2004) predictions for episodes of mortality seems to be reasonably well as it can be

depicted from the predicted plot. PACF plot shows that mortality is high during first three

days at lag0 and after that it starts declining. Residual plot shows that difference between

actual predicted values is evenly distributed around zero. For Lag0 the day of week terms

(Monday through Sunday) were highly significant (p < .00001), along with relative humidity

(p < 0.05) and wind speed (p < 0.05).

-20

24

Day

Re

sid

ua

l


0 5 10 15 20 25 30

-0.1

00

.00

0.1

0

Lag

Pa

rtia

l A

CF

10

20

30

40

50

Day

Nu

mb

er

of

ep

iso

de

s



mortality (Lag1)



2331

We observe almost the similar pattern for Lag1 with significant effects of temperature on

mortality for at least 2nd to 4th day (PACF plot). The predicted model fits well to the observed

data (Figure 6). The smoothers of day effect (p < .00001) and relative humidity (p <.05) were

significant for Lag1, as well as for wind speed (p < 0.05).

-20

24

Day

Re

sid

ua

l


0 5 10 15 20 25 30

-0.1

00

.00

0.1

0

Lag

Pa

rtia

l A

CF

10

20

30

40

50

Day

Nu

mb

er

of

ep

iso

de

s



mortality (Lag2)

Very little difference was observed between Lag2 and Lag1 in terms of PACF, residual and

predicted plots (Figure 7). The smoothers for day effect (p < .00001), relative humidity (p <

0.0001) and wind speed (p <. 05) were found to be significant.



2332

-4-2

02

4

Day

Re

sid

ua

l


0 5 10 15 20 25 30

-0.1

00

.00

0.1

0

Lag

Pa

rtia

l A

CF

10

20

30

40

50

Day

Nu

mb

er

of e

pis

od

es



mortality (Lag3)

For Lag 3, concentrated band was observed between 20-32 (number of episodes of mortality,

Figure 8) and smoothers for day effect (p < .00001), relative humidity (p < 0.0001) were

significant but non-significant for wind speed (p > 0.05). Temperature and mortality plot

(Figure 9) shows that there are more deaths during summer (> 30 0C) season as compared to

the winter (< 15 0C) season (all cause mortality). It is also evident that from Figure 10 that,

there are more male deaths as compared to female deaths.

3.3 Sensitivity analysis

Separate analysis was carried out for elderly people (age>65 years), however, smaller number

of deaths in 0-4 year’s age group did not allow us to do age specific analysis for this group.

There are more deaths for elder people (> 65 year) during winter season, particularly, when

the temperature is below 15 0C (Figure 11) as compared to summer season. All the Lag

effects except Lag3 were found to be significant (Table 4). For Lag0, Lag1, Lag2 smoothers



2333

were found to be significant for day effect (p < .00001), humidity and wind speed (p < .05).

However, for Lag3, significant smoothers were observed for day effect (p <. 00001) as well

as for relative humidity (p < .01) but non-significant for wind speed (p > .05).

0 200 400 600 800 1000

010

20

30

40

50

Days

Num

ber of D

eath

s

Total DeathsMale DeathsFemale Deaths

Figure 9 and 10: Temperature and Mortality Loess Plot and Daily Number of Deaths (2002-

04)

Figure 11: Temperature and Mortality Plot (>=65 years)



2334

Table 4: Sensitivity Analysis for Elderly People (age≥65 Years)

Penalized Spline Model: PS(8,4,4)

Mean temp 95% CI

Lag

Effects N β-Coeff.

Relative


Lag 0 1096 0.013014 1.13898783 0.0195 ** 1.021329 1.270201

Lag 1 1095 0.015657 1.16949262 0.000661 * 1.069000 1.279432

Lag 2 1094 0.010214 1.10753852 0.0205 * 1.016007 1.207316

Lag 3 1093 0.007907 1.08228008 0.0687 0.994024 1.178372 * (p<.01), ** (p<.05)

Table 5: Sensitivity Analysis (Female)


Mean temp 95% CI

Lag

Effects N β-Coeff.

Relative


Lag 0 1096 0.018712 1.205771969 0.00117* 1.349536 1.077323

Lag 1 1095 0.019845 1.21951105 0.0000304* 1.799326 0.826536

Lag 2 1094 0.010412 1.109733615 0.0213** 1.212415 1.015749

Lag 3 1093 0.013216 1.141290911 0.00316* 1.245719 1.045617

**(p<.01),**(p<.05)

Table 6: Sensitivity Analysis (Male)


Mean temp 95% CI

Lag

Effects N β-Coeff.

Relative


Lag 0 1096 0.012192 1.129663725 0.00474* 1.229192 1.038195

Lag 1 1095 0.00622 1.064175158 0.078 1.140321 0.993114

Lag 2 1094 0.005259 1.053997418 0.12 1.126077 0.986532

Lag 3 1093 0.00032 1.003203119 0.923 1.069985 0.940589 *(p<.01)

In order to identify the gender sensitivity of the model, separate analysis was carried out for

male and female deaths. Lag0, Lag1, Lag2 and Lag3 effects were significant for females but

for male it was only significant for Lag0. For every 10°C increase in temperature, mortality

for elderly people increased by 1.14%, for female by 1.21% and for males by 1.13%. Also for

every 10°C decreased (below 25°C), mortality increased by 0.88% for elderly, 0.83% for

females and 0.89% for males.

4. Results

The estimated number of deaths in Ludhiana city varies from 9000 to 10000 deaths mid-year

population. The total recorded deaths in Ludhiana city from 2002 – 2004 were 28,007 with an

average of 25.4 deaths per day. If the deaths of non-residents were excluded, the total deaths

were 19335 (69.3%). The age and sex distribution of deaths is shown in Table 2. There were



2335

more deaths among males (65%) and in the above 45 years age group (67%). Only 787

(2.8%) of the deaths were due to accidents and they have been excluded from the analysis.

Most of the deaths (71.2%) fall under the category of “symptoms, signs and abnormalities

classified elsewhere” due to improper certification of the cause of death. The mean (s.d.) of

temperature was 25.6(7.9)°C, relative humidity 58.1(19.3)%, and wind speed 6.87(4.88)

km/hour. The annual mean temperature ranged from 8.1°C to 41.5°C for 2002, 6.2°C to

41.5°C for 2003 and from 8.4°C to 40.6°C for 2004; relative humidity from 11.0% to 93.7%

for 2002, 10.3% to 99.7% for 2003 and from 13.7% to 99.7% for 2004 and wind speed varies

from 0.00 to 42.67 for 2002, 0.00 to 35.67 for 2003 from 0.00 to 27.33 for 2004.

We have worked out the PACF and Residual plots for Lag0, Lag1, Lag2 and Lag3 along with

predicted plot of mortality in R using, PS (8,4,4) model which is having the least GCV score.

It is evident from PACF and Residual plots that there is significant effect of temperature on

mortality.All the Lag effects except Lag3 (Table 4) were found to be significant. For Lag0,

Lag1, Lag2, smoothers were found to be significant for day effect (p < .00001), humidity and

wind speed (p < .05). However, for Lag3, significant smoothers were observed for day effect

(p < .00001) as well as for relative humidity (p < .01) but non-significant for wind speed (p >

.05).

The association between temperature and mortality by smoothing the effects of day of the

week, relative humidity and wind speed was found to be statistically significant. For every

10°C increase in temperature, mortality due to natural causes increased by 1.17% and for

every 10°C decrease in temperature (below 25°C), mortality increased by 0.86%. Sensitivity

analysis was carried out separately for deaths in the 65+ year’s age group and gender. Smaller

number of deaths in 0-4 year’s age group did not allow us to do age specific analysis.

Sensitivity analysis shows that elderly (> 65 years) population was much affected by

temperature variations. For every 10°C increase in temperature, mortality for elderly people

increased by 1.14%, for female by 1.21% and for males by 1.13%. Also for every 10°C

decreased (below 25°C), mortality increased by 0.88% for elderly, 0.83% for females and

0.89% for males.

5. Discussion

As a part of the Health Effects Institute, Boston USA, project, this paper gives the impact of

temperature on mortality in Ludhiana City of India. The mortality rates of Urban Punjab for

the years 2002, 2003 and 2004 were 6.2, 6.0 and 7.2 deaths per 1,000 mid-year population

(Source: SRS Bulletin). Apart from the analysis of the relation between temperature and

mortality in the city, the smoothing effects of day, relative humidity and wind speed have

also been worked out. The PACF i.e. partial autocorrelation function and residual plots were

plotted for Lag0, Lag1, Lag2 and Lag3 along with the predicted plots (all cause mortality). It

is evident from PACF and Residual plots that there is significant effect of temperature on

mortality after smoothing the effects of day, relative humidity and wind speed.

Sensitivity analysis was carried out separately for deaths in the 65+ year’s age group as well

for gender. Smaller number of deaths in 0-4 year’s age group did not allow us to do age

specific analysis. Sensitivity analysis shows that elderly (> 65 years) population was much

affected by temperature variations, particularly, during winter season. Studies based on

retrospective data face several difficulties and this study is no exception. On the basis of the

age-specific mortality rates and age distribution of the population of urban India, it seems that

about 15-23% may not have been registered in various age groups, however, there is no



2336

suggestion that this proportion would vary day by day. In conclusion there is need to improve

the overall registration system of deaths. Cause specific analysis was not possible as cause of

death was not clearly mentioned in most of the cases. It seems there is under reporting of

deaths, particularly, among children below the age of 5 years. We suggest that ICD-10 coding

system should be introduced properly so that each case should have proper code and cause-

specific analysis may be performed to get better results in future.

Acknowledgments

We are thankful to Birth and Death Registration Office, Ludhiana; Meteorological

Department, New Delhi and Health Effects Institute, Boston.

6. References

1. Anderson B.G., and Bell M.L., (2009), Weather-related mortality: how heat, cold, and

heat waves affect mortality in the United States. Epidemiology, 20(2), pp 205-213.

2. Baker DANM, (2008), Environmental Epidemiology: Study Methods and Application,

Oxford University Press.

3. Ballester F., Corella D., Perez-Hoyos S., Saez M.and Hervas A.,(1997), Mortality as a

function of temperature. A study in Valencia, Spain, 1991 1993, International Journal

ofJ Epidemiology, 26(3),pp 551-561.

4. Ballester J., Robine J.M., Herrmann F.R. and Rodo X, (2011), Long-term projections

and acclimatization scenarios of temperature-related mortality in Europe, Nat

Communications , 2, pp 358.

5. Barron D.N., (1992), The Analysis of Count Data: Overdispersion and

Autocorrelation, Social Methodology, 22, pp 179-220.

6. Basu, R.,(2009), High ambient temperature and mortality: A review of epidemiologic

studies from 2001 to 2008. Environmental Health, doi:10.1186/1476-069X-8-40.

7. Basu R. and Malig B, (2011), High ambient temperature and mortality in California:

Exploringthe roles of age, disease, and mortality displacement. Environmental

Research, 111(8), pp 1286-1292.

8. Braga A.L., Zanobetti A. and Schwartz J. (2001), The time course of weather related

deaths, Epidemiology, 12, pp 662–667.

9. Carson C., Hajat S., Armstrong B.,and Wilkinson P, 2006, Declining vulnerability to

temperature related mortality in London over the 20th century, American Journal of

Epidemiology, 164(1), pp 77-84.

10. Curriero F.C., Heiner K.S., Samet J.M., Zeger S.L., Strug L. and Patz J.A.,2002,

Temperature and mortality in 11 cities of the eastern United States, American Journal

of Epidemiology, 155(1), pp 80-87.

11. Ellis F.P., and Nelson F.,(1978), Mortality in the elderly in a heat wave in New York

City, Environmental Research, 15, pp 504-512.



2337

12. Fouillet A., Rey G. And Laurent F, (2006), Excess mortality related to the August

2003 heat wave in France, International Archives of Occupational and Environmental

Health, 80,pp 16-24.

13. Goldberg M.S., Gasparrini A., Armstrong B. and Valois M.F., (2011), The short-term

influence of temperature on daily mortality in the temperate climate of Montreal,

Canada. Environmental Research, 111(6), pp 853-860.

14. Hajat, S., Armstrong B., Baccini M., Biggeri A., Bisanti L., Russo A., Paldy A.,

Menne B.and Kosatsky T., 2006, Impact of high temperatures on mortality: is there an

added “heat wave” effect? Epidemiology, 17, pp 632-638.

15. Hajat S., Kovats R.S., Atkinson R.W. and Haines A.,(2002), Impact of hot

temperatures on death in London: a time series approach. J Epidemiol Community

Health, 56(5), pp 367-372.

16. Ha J., Shin Y. and Kim H.,(2011), Distributed lag effects in the relationship between

temperature and mortality in three major cities in South Korea, Science of the Total

Environment ,409(18), pp 3274-3280.

17. Hoffmann B., Hertel S., Boes T., Weiland D. and Jockel K.H.:, (2008), Increased

cause-specific mortality associated with 2003 heat wave in Essen, Germany, Journal

of Toxicology and Environmental Health, 71(11-12), pp 759-765.

18. Huynen, M.M., Martens, P., Schram, D.,Weijenberg, M.P. and Kunst, A.E.,(2001),

The impact of heat waves and cold spells on mortality rates in the Dutch population.

Environmental Health Perspectives, 109, pp 463-470.

19. Kim H., Ha J.S. and Park J. ,(2006), High temperature, heat index, and mortality in 6

major cities in South Korea, International Archives of Occupational and

Environmental Health, 61(6), pp 265-270.

20. Kumar R., Sharma S., Thakur J.S., Lakshmi P.V.M., Sharma M.K. and Singh T.,

(2010), Association of air pollution and mortality in the Ludhiana City of India: A

time Series Study, Indian Journal of Public Health , 54( 2) , pp 98-103.

21. McKee C.M.,(1989), Deaths in winter: can Britain learn from Europe? European

Journal of Epidemiology, 5(2), pp178-182.

22. Muggeo V.M., (2008), Modeling temperature effects on mortality: multiple

segmentedrelationships with common break points, Biostatistics, 9(4), pp 613-620.

23. Ostro B.D., Roth L.A., Green R.S. and Basu R, (2009), Estimating the mortality effect

of the July 2006 California heat wave, Environmental Research , 109(5), pp 614-619.

24. Peng R.D., Dominici F. and Louis T.A., (2006), Model Choice in Time Series Studies

of Air Pollution and Mortality, Journal of the Royal Statistical Society Part A, 169(2),

pp 179-203.

25. Schwartz J., (2000), The distributed lag between air pollution and daily deaths,

Epidemiology, 11(3), pp 320- 326.



2338

26. Yu W., Guo Y., Ye X., Wang X., Huang C., Pan X. and Tong S., (2011),The effect of

various temperature indicators on different mortality categories in a subtropical city of

Brisbane, Australia, Science of the Total Environment , 409(18), pp 3431-3437.

27. Hastie T.J. and Tibshirani R.J., (1990), Generalized Additive Models,Chapman and

Hall/CRC, London.

28. Peng R.D. and Dominici F., (2008), Statistical methods for environmental

epidemiology in R, Springer, NewYork .

29. Wood S.N.,(2006), Generalized Additive Model: an introduction with R, Chapman

and Hall/CRC, New York.

Documents

The impact of temperature on mortality in Ludhiana … impact of temperature on mortality in Ludhiana City, India: A time series analysis Suresh Kumar Sharma et al International Journal