Airborne particulate matter and gaseous air pollutants in

residential structures in Lodi province, Italy

Introduction

In urban and suburban environments, the pollutantsof main concern are PM, nitrogen dioxide (NO2),carbon monoxide (CO), and ozone (O3), because oftheir widely documented health impacts (Dockeryet al., 1993; Pope, 2000; Weschler, 2006). However,some of these pollutants are produced both by

outdoor sources such as traffic exhaust and by indoorsources of combustion (Abt et al., 2000; Kousa et al.,2001). Some recent epidemiological studies examinethe health effects of personal exposure to PM,differentiated by outdoor (ambient) or indoor (non-ambient) origin, finding that markers of airwayinflammation (Koenig et al., 2005) and cardio-respi-ratory adverse effect estimates (Ebelt et al., 2005) are

Abstract The province of Lodi is located in northern Italy on the Po River plain,where high background levels of air pollutants are prevalent. Lodi province ischaracterized by intensive agriculture, notably animal husbandry. This paperassesses indoor levels of selected airborne pollutants in 60 homes in the province,with special attention to size-fractionated particulate matter (PM). Indoor PM2.5

concentrations are frequently higher than current guidelines. PM10 and nitrogendioxide also exceed the respective guideline recommendations in some cases,noting that 24-h nitrogen dioxide levels were compared with an annual limitvalue. All other studied pollutant levels are below current international guide-lines. Among indoor PM size fractions, PM0.5 is predominant in terms of massconcentrations corresponding to 57% of PM10 in summer and 71% in winter.A strong seasonal trend is observed for all studied pollutants, with higher levelsin winter corresponding to changes in ambient concentrations. The seasonalvariation in PM10 is largely due to PM0.5 increase from summer to winter.Summer indoor PM levels are mainly from indoor-generated particles, whileparticles of outdoor origin represent the main contribution to winter indoorPM levels. On average, indoor concentrations of coarse PM are mostlyconstituted by indoor-generated particles.

A. Cattaneo1,2, C. Peruzzo3,G. Garramone4, P. Urso1,R. Ruggeri5, P. Carrer1,D. M. Cavallo6

1Department of Occupational and EnvironmentalHealth, Universit� degli Studi di Milano, Milano, Italy,2Unit of Epidemiology, Fondazione IRCCS Ca�Granda-Ospedale Maggiore Policlinico, Milan, Italy,3Occupational Health Unit, Macchi Foundation Hospital,Varese, Italy, 4International Centre for Pesticides andHealth Risk Prevention, L. Sacco Hospital, Milano, Italy,5Environmental Protection Agency of Lombardy Region(ARPA), Sondrio, Italy, 6Department of Chemical andEnvironmental Sciences, Universit� degli Studidell�Insubria, Como, Italy

Key words: Size-fractionated particulate matter; Nitro-gen dioxide; Carbon monoxide; Ozone; Carbon dioxide;Residential homes.

A. CattaneoDepartment of Occupational and Environmental HealthUniversit� degli Studi di MilanoVia San Barnaba, 820122 MilanoItalyTel.: +39 02 503 20147Fax: +39 02 503 20111e-mail: andrea.cattaneo@unimi.it

Received for review 28 April 2011. Accepted forpublication 17 June 2011.

Practical ImplicationsThis study presents a comparison between measured indoor concentrations in the study area and indoor air qualityguideline criteria. Accordingly, particulate matter (PM) and NO2 are identified as key pollutants that may pose healthconcerns. It is also found that indoor PM in residential units is mainly constituted by particles with aerodynamicdiameters <0.5 lm, especially in winter. Risk mitigation strategies should be focused on the reduction in indoor levelsof NO2 and ultrafine and fine particles, both infiltrated from outdoors and generated by indoor sources.

Indoor Air 2011; 21: 489–500wileyonlinelibrary.com/journal/inaPrinted in Singapore. All rights reserved

� 2011 John Wiley & Sons A/S

INDOOR AIRdoi:10.1111/j.1600-0668.2011.00731.x

489

more influenced by outdoor-generated PM. On thecontrary, there is evidence that indoor PM2.5 levelsare responsible for adverse health effects mainlyassociated with exposure to environmental tobaccosmoke (Osman et al., 2007). For CO, indoor concen-trations can be much higher than outdoor levels,leading in extreme cases to lethal consequences.Consequently, it is evident that the measurement ofindoor air pollutant concentrations is important tobetter estimate health risks and that understandingindoor exposure determinants is critical to the devel-opment of specific risk management plans.Human exposure to air pollutants can be directly

measured using personal and stationary (indoors, out-doors, at ambient level) exposure measurement strate-gies or direct measurements, or they can be estimatedusing models that predict exposure from surrogates(Zou et al., 2009). In general, ambient measurementsmay be used as proxies of personal exposure to fine PM(Brown et al., 2008), depending on the small-scalespatial variability of the studied pollutants; however,several studies showed that the best indicator ofpersonal exposure is represented by indoor concentra-tions (Adgate et al., 2003; Brown et al., 2008; Janssenet al., 2005), as people spendmost of their time in indoorenvironments. Nevertheless, it should be noted thatultrafine PM is mostly constituted by freshly emittedparticles, which are generally characterized by highspatial variability. Thus, the discrepancies betweenpersonal and indoor ultrafine PMconcentrations shouldincrease compared with other PM-size fractions (Siou-tas et al., 2005). Factors that contribute to the differencebetween ambient/outdoor and personal/indoor airpollution levels include human activities, exposure totobacco smoke, ventilation rate, and infiltrationefficiency (Brown et al., 2009; Sarnat et al., 2000).This study aims to measure and evaluate indoor

concentrations of size-fractionated PM with a 50%cut-off aerodynamic diameter (Da) of 10 (PM10), 2.5(PM2.5), 1.0 (PM1), 0.5 (PM0.5), and 0.25 (PM0.25) lm,along with NO2, O3, CO, and CO2 in residentialdwellings located in the province of Lodi. AlthoughPM fractions with Da <1 lm are of concern becauseof their potential effects on health (Delfino et al.,2005), they are not usually investigated in indoorenvironments.Lodi province is located in the Po Valley, which has

high population densities and high emissions of pollu-tants. This area is also characterized by high atmo-spheric stability caused by low wind speeds andtemperature inversions, especially in winter (Vecchiet al., 2004). Indoor levels of pollutants of healthconcern measured in the Po Valley are generally higherthan average concentrations measured in residentialstructures across Europe. On average, indoor levels ofPM2.5, NO2, and CO measured in dwellings in Milan(about 40 km from Lodi) during the EXPOLIS study

were 42.7, 82.9 lg/m3, and 1.9 ppm, respectively(Bruinen De Bruin et al., 2004; Rotko et al., 2002).PM2.5 and CO levels in Milan were the highest amongthe six European cities studied (Georgoulis et al., 2002;Rotko et al., 2002), while NO2 concentrations weremarkedly higher than those measured in Helsinki,Basel, Oxford, and Prague and slightly lower thanindoor levels measured in Athens (Rotko et al., 2002).

Materials and methods

Study design

Air samples were collected for about 24 h (meansampling time 1409 min) in the living rooms of 60homes located in the province of Lodi (Figure 1).These homes were selected from about 260 dwellings inthe study area to ensure adequate representation of thevarious building types. The following factors wereconsidered when selecting dwellings for inclusion inthis study: dwelling type (single- or multifamily),location (urban or rural area), proximity to potentialPM sources (roads, point emission sources), windowtype, ventilation and heating system type (natural ormechanically ventilated, indoor/outdoor gas boilers,wood stoves).Indoor levels of pollutants were measured for 24 h in

each of the 60 selected homes by a team of two trainedtechnicians from May through September 2007 (�sum-mer� – heating system switched off). The same samplingprocedure was repeated in 53 of the 60 homes fromDecember 2007 through March 2008 (�winter� – heatingsystem switched on). The indoor monitoring wasperformed over the first 4 days of the week (Mon-day–Thursday), leaving Friday for calibration and data

Fig. 1 Map of the province of Lodi, showing the location of thehomes selected for the study (white drops), the position of fixedsampling stations (black triangles), and the position of thelargest cities in the area (>10 000 inhabitants) (white circles)

Cattaneo et al.

490

storage procedures. Because of the limited number ofinstruments available, a maximum of two homes weresimultaneously monitored each day, so that 2–8residential dwellings were monitored each weekdepending also on the availability of householders.

Instrumentation

Particulate matter and gaseous pollutants were mon-itored indoors using two multipollutant monitoringstations (MMS). MMS (Figure 2) are composed ofportable, sound-absorbent containers made of med-ium-density fiberboard (Koistinen et al., 1999) modi-fied to collect size-fractionated PM (PM10, PM2.5, PM1,PM0.5, and PM0.25), NO2, and O3 by indirect methods,as well as CO and CO2 using direct-reading instru-ments. Ambient concentrations were simultaneouslycollected by the Regional Environmental ProtectionAgency of Lombardy, using several fixed stationsbelonging to the regional air quality network.Indoor PM measurements were performed using

low-flow sampling pumps calibrated with a primarystandard soap bubble meter (mini-Buck Calibrator; APBuck Inc., Orlando, FL, USA), following the factorysuggested protocols. PM10 and PM2.5 were sampledusing GK2.69 and GK2.05 (BGI Inc, Waltham, MA,USA) cyclones, respectively. GK2.69 was connected toZambelli Ego personal sampling pumps (Zambelli Srl,Bareggio, MI, Italy) with flow rates of 1.6 l/min, andGK2.05 was coupled with Zambelli Chronos personalsampling pumps calibrated at 4.0 l/min. The samplingpumps were connected to AC power supply to extendthe sampling time to 24 h.Different PM fractions were collected via personal

cascade impactor sampler (PCIS) (SKC Inc, Eighty

Four, PA, USA) (Misra et al., 2002; Singh et al., 2003),which allow high-volume sampling and proportionalPM mass collection on filters, with the advantage ofgreater accuracy of gravimetric determination withregard to the finest size fraction.PM10, PM2.5, and PM0.25 were collected on 37-mm,

2.0-lm polytetrafluoro-ethylene PTFE filters withpolymethylpentene support ring, and PCIS impactionstages were constituted via 25-mm, 0.5-lm PTFEmembranes (SKC Inc). Because of the greater collectedmass, more complete dataset, and additional informa-tion about size fractions, only PM concentrationscollected by PCIS were used in indoor data analysis.PM mass was determined by gravimetric analysis. Thenet PM mass on filters was measured by weighing theconditioned filters before and after sampling with amicrobalance in a temperature (20 ± 1�C)- and rela-tive humidity (50 ± 5%)–controlled environment(Activa Climatic; Aquaria, Lacchiarella, MI, Italy).The quality of the weighing procedure was assessedusing the ASTM D 6552 method [American Society ofTesting and Materials (ASTM), 2000]. A mass limit ofdetection (LOD) of 7 lg (weighing imprecision =18 lg) for 37-mm filters and 5 lg (weighing impreci-sion = 10 lg) for 25-mm filters was calculated.Consequently, the mean LOD of PM concentrationssampled with PCIS was 0.6 lg/m3 for PM0.25, 1.0 lg/m3 for PM0.5, 1.4 lg/m3 for PM1, 1.8 lg/m3 for PM2.5,and 2.2 lg/m3 for PM10. The LOD of PM10 concen-trations collected via GK2.69 was 3.1 lg/m3 and thatof PM2.5 sampled via GK2.05 was 1.2 lg/m3.Agreement between PCIS and GK samplers was

evaluated according to the indications summarized byWatson et al. (1998). Equivalence between the twosampling methods was defined as a function of slope

Fig. 2 The multipollutant monitoring station. (a) Front view, with particulate matter separators and the opened sampling pumpcontainer. (b) Rear view, showing all the external instruments and sampling probes

Airborne particulate matter and gaseous air pollutants

491

[1 ± 3 standard errors (s.e.)], intercept (0 ± 3 s.e.),and Pearson coefficient (r) (>0.9). If the r criterion wasfulfilled but slope and intercept criteria were not met,the compared methods were deemed characterized byreciprocal predictability. Data with r < 0.9 are classi-fied as not comparable.Indoor NO2 and O3 were collected using radial

symmetry diffusive samplers (Radiello; Sigma-AldrichInc., Milano, Italy) (Cocheo et al., 1996) and analyzedvia ultraviolet-visible spectrophotometry at a wave-length of 537 and 430 nm, respectively, using a doublebeam spectrophotometer (Cary 100; Varian Inc., PaloAlto, CA, USA) and following the factory operationalprotocols.Carbon monoxide concentrations were continuously

collected to ensure correct analysis of short-termindoor exposures. Passive T15v Langan electrochemi-cal sensors (Langan Products Inc, San Francisco, CA,USA) were used for this purpose. These instrumentswere used in an extensive European study on humanexposure in life environments (Georgoulis et al., 2002).Zeroing and calibrating procedures were carried outbefore and after the seasonal sampling sessions in glovebags at 20�C, using the climatic cabinet and twocertified standard gas mixtures containing <0.5 and10.3 ppm of CO in air. Twenty-four-hour concentra-tions of CO were a posteriori corrected using thecorresponding average temperature, because of theknown interference of temperature on the instrumentalbaseline.CO2 levels were determined indoors and outdoors

with Telaire 7000 dual path non-dispersive infraredanalyzers cabled to HOBO H8 data loggers (OnsetComputer Inc., Pocasset, MA, USA), which were alsoused to measure temperature (T) and relative humidity(rH). CO, CO2, T, and rH sampling frequencies wereset to 1 min. Zeroing of CO2 analyzers was performedbefore each seasonal campaign.The Regional Agency for Prevention and Environ-

ment of Lombardy routinely monitors atmosphericpollution levels using a network of fixed monitoringstations, with 2-h averaged daily measurements madepublic on a day-to-day basis. For the purposes of thisstudy, we used PM2.5, PM10, O3, NO2, and CO ambientconcentrations sampled at the nearest monitoringstation in the same period over which indoor measure-ments were performed.

Analysis

Indoor–outdoor air exchange rates (AERs) at each sitewere estimated from the decay of indoor CO concen-trations after emission peaks, using the methoddescribed in Polidori et al. (2007).The particulate infiltration factor (FINF) and the

concentration of indoor-generated particles (Cig) wereestimated from the regression of indoor concentrations

(Ci) against ambient concentrations (Ca), using thefollowing equation (Hanninen et al., 2004):

Ci ¼ FINFCa þ Cig ¼ Cog þ Cig: ð1Þ

The slope of the regression estimates the mean FINF,and the intercept estimates the average concentrationof indoor-generated PM (Cig), as demonstrated by Ottet al. (2000). Moreover, we estimated seasonal out-door-generated PM (Cog) indoors from the calculatedFINF and the corresponding mean seasonal ambientlevels.Statistical data analysis was carried out using the

SPSS package (SPSS Inc., Chicago, IL, USA). PMsamplers were compared by the Bland–Altman plotsmethod (Altman and Bland, 1983; Bland and Alt-man, 1986). Descriptive statistics were performed onpollutants, including pollutant cumulative distribu-tion curves and histograms of seasonal PM fractioncontributions. Gaussian distribution of parameterswas verified via a Kolmogorov–Smirnov test. Differ-ences between summer and winter monitoring werestudied by paired t-test if parameters showed normalor log-normal distribution and by Wilcoxon testif neither log-normal distribution was achieved.Linear regression models for single monitoring sessionsand linear mixed models for repeated measurementswere carried out to examine the indoor–outdoorrelationship. Significance level was established tobe <0.05.

Results

Principal characteristics of the 53 residential units twicemonitored in the study are summarized in Table 1. Theselected sample of dwellings is characterized by inhab-itation by ‡2 people in >90% of cases and aprevalence of non-smoking households. Sample resi-dential units are homogeneously distributed betweensingle- and multifamily houses and are located mainlyin suburban and urban areas, near streets with low-automotive traffic (70% facing streets with <5 vehi-cles/min) and at ground, first, or second floors (89% ofcases).

PM sampler comparison

PM2.5 and PM10 indoor measurements simulta-neously collected with the multistage PCIS impactorand GK cyclones were first compared by regressionanalysis of paired concentrations. The PM2.5 regres-sion line presented slope = 1.04 (ES = 0.050), inter-cept = 9.4 lg/m3 (ES 1.8 lg/m3), and r = 0.92. Theregression analysis of PM10 paired data showed aslope of 0.92 (ES 0.034) and an intercept of 6.7lg/m3 (ES 1.6 lg/m3), with r = 0.95. In addition,Bland–Altman plots were depicted to better assess

Cattaneo et al.

492

absolute errors and possible error trends and toevaluate these errors as a function of desiredprecision (Figure 3). It should be noted that the95% limits of agreement were generally quite large(about ±10 lg/m3), especially for PM2.5 measure-ments. Moreover, PM2.5 concentrations collected byGK2.05 cyclone were significantly lower than thosemeasured using PCIS impactor: the absolute differ-ence in single measurements was up to )80 lg/m3,leading to an average difference of )10.4 lg/m3

(Figure 3a). Thus, PM2.5 levels measured by PCISand GK2.05 denoted inadequate agreement betweenthe two testing methods. PM10 measurements showedbetter overall agreement, with a mean absolutedifference of )3.6 lg/m3 and less pronounced dis-agreement trends.

Indoor and ambient pollutant levels

As expected, indoor and ambient air pollutant levelswere lower in summer than in winter (Tables 2 and 3),with the exception of O3. On average, PM10 andPM2.5 concentrations increased indoors by 23.5 and24.7 lg/m3, respectively, from summer to winter,while NO2 and CO indoor levels increased by19.6 lg/m3 and 0.3 ppm, respectively. Such seasonaldifferences were statistically significant for all consid-ered pollutants (P < 0.05). In addition, estimatedAER increased by 2.5 times from winter to summer(Table 2).

Comparison with international and national guidance values

Recently, the World Health Organization (WHO)developed specific guidance values (GV) (WHO,2010) for selected indoor pollutants, including CO.Other air quality guidelines (AQG) may be taken intoaccount to evaluate atmospheric concentrations ofthe other studied pollutants in all non-occupationalenvironments (WHO, 2006) (Table 4).Indoor PM2.5 levels exceeded (45.9%) WHO AQG

limits in several cases, especially during the winterperiod (71.7%). WHO AQG levels were also exceededby 19.2% for indoor measurements of total PM10

(5.8% among measurements performed during summerand 34.0% during winter) (Figure 4).Measured levels of indoor gaseous pollutants are not

strictly comparable with WHO guidelines, because ofdiffering average exposure times; however, it is reason-able to assume that they did not exceed WHO

Table 1 Principal demographic and household characteristics of the 53 twice monitoredhouses

No. (%)

No. inhabitants1 5 (9.4)2 15 (28.3)3 17 (32.1)‡4 16 (30.2)

Smoking habitsNo 37 (69.8)Yes 16 (30.2)

Type of houseFlat 18 (34.0)Independent house 35 (66.0)

Traffic density of neighboring streets (vehicles/min)>20 7 (13.2)5–20 9 (17.0)1–5 24 (45.3)<1 13 (24.5)

Location areaTown 20 (37.7)Suburb 26 (49.1)Rural area 7 (13.2)

Floor)1 2 (3.8)0 18 (34.0)1 15 (28.3)2 13 (24.5)‡3 4 (7.5)

Fig. 3 Comparison of PM2.5 (a) and PM10 (b) sampled with GKcyclones and with personal cascade impactor sampler impactorusing Bland–Altman plots. Solid lines represent the observedaverages, whereas broken lines represent the 95% limits ofagreement

Airborne particulate matter and gaseous air pollutants

493

recommended levels for O3 (100 lg/m3 8-h meanexposure), while we cannot exclude some levels exceed-ing of 200 lg/m3 1-h mean exposure limit for NO2. Ifwe consider mean values of 24-h NO2 sampling insummer and winter as a rough approximation of theannual mean (worst and best cases), then the 40 lg/m3

WHO annual limit would have been exceeded in 19.5%of cases.Twenty-four-hour concentrations of indoor CO were

first compared with the WHO 24-h recommended limit;this was never exceeded. As CO data were continuouslycollected, 15-, 30-min, 1-, and 8-h averages werecalculated and compared with corresponding GV;these were never exceeded.Indoor CO2 measures exceeded ASHRAE-recom-

mended indoor levels (1000 ppm) [American Society ofHeating Refrigerating and Air-Conditioning Engineers(ASHRAE), 1989] in 12.2% of cases, mainly in winter

Table 2 Descriptive statistics for size-fractionated PM, gaseous pollutants (O3, NO2, CO),and CO2 indoor concentrations. Indoor microclimatic parameters (T, rH%), AER, and I/Oratios are also reported

n AM € s.d.Median[25th–75th] 95th S vs. W

PM0.25 (lg/m3)Summer 55 11.4 € 11.3 8.7 [7.1–11.7] 40.6 *Winter 49 24.3 € 22.6 16.3 [11.0–27.9] 88.6Total 104 17.5 € 18.6 11.0 [7.4–18.3] 68.3

PM0.5 (lg/m3)Summer 54 16.8 € 13.6 14.7 [10.1–18.0] 48.7 *Winter 48 37.6 € 32.9 27.8 [18.3–41.2] 122.9Total 102 26.6 € 26.6 17.8 [12.0–28.9] 106.9

PM1 (lg/m3)Summer 52 18.5 € 14.0 16.4 [11.6–20.5] 50.4 *Winter 47 41.9 € 35.4 31.3 [21.0–46.1] 130.0Total 99 29.6 € 28.8 20.3 [13.5–36.5] 124.8

PM2.5 (lg/m3)Summer 52 21.6 € 14.1 19.4 [14.6–23.9] 53.1 *Winter 46 46.3 € 36.9 36.0 [23.0–55.3] 138.2Total 98 33.2 € 29.8 23.0 [16.5–36.5] 129.1

PM10 (lg/m3)Summer 52 29.7 € 15.0 26.8 [21.3–34.6] 61.7 *Winter 47 53.2 € 37.5 42.0 [28.2–63.6] 143.7Total 99 40.9 € 30.3 32.8 [22.8–44.4] 132.7

O3 (lg/m3)Summer 60 12.3 € 11.5 8.7 [3.0–17.1] 37.0 –Winter 53 n.d. n.d. n.d.Total – – – –

NO2 (lg/m3)Summer 60 18.9 € 9.8 16.1 [11.7–25.1] 37.1 *Winter 53 38.5 € 16.8 37.2 [26.9–45.7] 64.1Total 113 28.1 € 16.7 25.3 [14.1–37.5] 59.2

CO (ppm)Summer 54 1.4 € 0.7 1.6 [0.8–1.9] 2.3 *Winter 50 1.7 € 0.8 1.5 [1.3–1.8] 3.6Total 104 1.5 € 0.8 1.5 [1.1–1.9] 3.3

CO2 (ppm)Summer 39 569.6 € 330.5 478.4 [451.3–545.5] 1127.6 **Winter 35 877.9 € 424.2 801.7 [638.9–959.9] 2000.7Total 74 715.4 € 405.8 594.9 [465.2–841.5] 1716.9

T (�C)Summer 54 25.5 € 2.1 25.3 [24.2–27.0] 29.8 *Winter 50 19.8 € 1.5 20.0 [18.8–20.8] 22.4Total 104 22.8 € 3.4 22.4 [20.0–25.4] 28.4

rH (%)Summer 39 45.5 € 9.4 45.1 [38.1–52.2] 63.7Winter 36 40.6 € 9.9 39.6 [34.3–44.9] 55.9Total 75 43.1 € 9.9 41.4 [36.8–46.4] 62.4

AER (/h)Summer 36 1.5 € 1.9 0.7 [0.4–1.9] 6.0 *Winter 27 0.6 € 0.8 0.3 [0.2–0.5] 3.0Total 63 1.1 € 1.7 0.5 [0.2–1.4] 5.1

I/O PM2.5

Summer 46 1.14 € 0.93 1.00 [0.71–1.20] 3.36 *Winter 44 1.30 € 0.90 1.04 [0.69–1.63] 3.88Total 90 1.22 € 0.91 1.03 [0.70–1.29] 3.84

I/O PM10

Summer 52 1.13 € 0.73 0.95 [0.80–1.22] 2.61 *Winter 47 0.93 € 0.78 0.61 [0.48–1.05] 2.68Total 99 1.04 € 0.76 0.86 [0.60–1.16] 2.63

AER, air exchange rates; CO, carbon monoxide; PM, particulate matter; n.d., not deter-mined.*Significant (P < 0.05) difference in summer vs. winter levels by t-paired test on normaland log-normal parameters. **Significant (P < 0.05) difference in summer vs. winterlevels by Wilcoxon test, if log-normal distribution was not verified.

Table 3 Descriptive statistics for PM2.5, PM10, O3, NO2, and CO ambient levels

n AM € s.d. Median [25th–75th] 95th S vs. W

PM2.5 (lg/m3)Summer 56 19.8 € 7.2 20.0 [16.0–25.3] 29.2 *Winter 51 38.3 € 20.3 32.0 [24.0–53.0] 79.8Total 107 28.6 € 17.6 23.0 [18.0–32.0] 70.0

PM10 (lg/m3)Summer 60 28.3 € 9.9 28.0 [22.3–34.0] 46.8 *Winter 53 65.2 € 30.8 58.0 [45.0–83.5] 127.6Total 113 45.6 € 28.8 34.0 [26.5–57.0] 114.5

O3 (lg/m3)Summer 60 57.1 € 17.9 60.5 [43.5–66.0] 93.0 *Winter 53 15.6 € 16.9 12.0 [4.0–21.5] 62.2Total 113 37.7 € 27.1 36.0 [12.5–61.5] 78.8

NO2 (lg/m3)Summer 60 26.7 € 11.4 24.0 [19.0–35.8] 51.9 *Winter 53 47.8 € 18.9 44.0 [32.0–56.5] 88.1Total 113 36.6 € 18.6 32.0 [23.0–46.0] 78.8

CO (ppm)Summer 60 0.49 € 0.2 0.5 [0.3–0.7] 0.8 *Winter 53 0.9 € 0.3 1.0 [0.7–1.2] 1.5Total 113 0.7 € 0.4 0.6 [0.4–1.0] 1.4

CO, carbon monoxide; PM, particulate matter.*Significant difference (P < 0.05) in summer vs. winter levels by t-paired test on normaland log-normal parameters.

Table 4 Recommended guideline values

InstitutionLimittype CO (ppm)a

CO2

(ppm)NO2

(lg/m3)O3

(lg/m3)PM10

(lg/m3)PM2.5

(lg/m3)

WHO IAQG 86 (15 min) 40 (1 year)30 (1 h)

8.6 (8 h)6.0 (24 h)

WHO AQG 86 (15 min) 40 (1 year) 100 (8 h) 50 (24 h) 25 (24 h)52 (30 min) 200 (24 h) 20 (1 year) 10 (1 year)

ASHRAE 1000

aConverted from values expressed as milligram per cubic meter, assuming an ambienttemperature of 20�C and an atmospheric pressure of 1013.25 hPa.AQG, air quality guidelines; PM, particulate matter.

Cattaneo et al.

494

(20.0%), while in summer, they exceeded the recom-mended limit in only 5.1% of total measurements.Mean mass contributions of PM size fractions to

indoor PM10 are depicted in Figure 5. A notablecontribution from the finest particle sizes was detected,such that PM0.5 represented more than 50% of indoorPM10 mass measured in both seasons. On average,indoor PM0.5 showed an important seasonal increase,rising from 16.7 lg/m3 (57% of PM10) in summer to37.7 lg/m3 (71% of PM10) in winter. An oppositepattern was observed for indoor coarse particles(Da = 2.5–10 lm), which decreased from 8.2 lg/m3

in summer to 6.9 lg/m3 in winter.

Indoor and ambient relationship

The influence of ambient pollutant concentrations onindoor levels is summarized in Table 5. Ambient PMconcentrations turned out to be significant predictorsof indoor concentrations, and the association betweenindoor and ambient pollutants was always positivewhen significant. Ambient concentrations explainedindoor levels with differing strengths, depending onseason and type of pollutant (at most 29% of totalvariability). The association between ambient andindoor PM2.5 was stronger during winter, while PM10

showed no seasonal trend. The correlation analysis onnon-log-transformed data allowed us to estimate thatan increment of 10% in PM10 and PM2.5 outdoor levelswas responsible for an increase of 4.8% and 4.1%,respectively, in indoor concentrations during summer,and for an increase of 5.6% and 6.9% during winter,respectively. On the whole, the same increase inoutdoor levels gave an indoor increment of 5.3% forPM10 and 7.4% for PM2.5, indicating a stronger

dependence of PM2.5 on outdoor levels with respectto PM10.With regard to O3, an increment of 10% in outdoor

concentrations was responsible for a 2.4% rise in indoorsummer levels. In summer, indoor NO2 concentrationswere weakly explained by outdoor levels, with a 2.5%indoor increase corresponding to an outdoor incrementof 10%; there was no significant association withambient concentrations during cold months. The whole

Fig. 5 Mean seasonal contributions of size-fractionated partic-ulate matter (PM) (PM2.5–10, PM1–2.5, PM0.5–1, PM0.25–0.5) toPM10 mean levels

Table 5 Simple linear regression between outdoor and indoor measurements for singlemonitoring session (�summer� and �winter�) and linear mixed model for total results.Ambient levels were considered as predictors of the corresponding indoor levels(dependent variables)

n R2Regressioncoefficient

Standarderror CI 95% t P

PM 2.5 (lg/m3)Summer 52 0.104 0.421 0.186 0.045 0.796 2.259 0.029Winter 46 0.287 0.702 0.171 0.358 1.047 4.112 0.000Total 54 – 0.745 0.110 0.526 0.964 6.764 0.000

PM 10 (lg/m3)Summer 52 0.217 0.495 0.133 0.228 0.762 3.723 0.001Winter 47 0.222 0.576 0.161 0.252 0.899 3.583 0.001Total 55 – 0.544 0.073 0.398 0.689 7.417 0.000

O3 (lg/m3)Summer 59 0.144 0.245 0.078 0.088 0.401 3.126 0.003Winter 53 n.d. n.d. n.d. n.d. n.d. n.d. n.d.Total – – – – – – – –

NO2 (lg/m3)Summer 59 0.090 0.259 0.108 0.042 0.475 3.704 0.020Winter 52 0.039 0.175 0.122 )0.070 0.421 1.436 0.157Total 60 – 0.454 0.073 0.309 0.599 6.205 0.000

CO (ppm)Summer 53 0.004 )0.199 0.449 )1.100 0.702 )0.443 0.660Winter 49 0.222 1.174 0.317 0.537 1.811 3.704 0.001Total 59 – 0.797 0.198 0.404 1.191 4.027 0.000

CO, carbon monoxide; PM, particulate matter; n.d., not determined.Determinant coefficient (R2), regression coefficient and standard error, 95% confidenceinterval, t-statistic (t) and P-value (P) are reported. PM10 and PM2.5 were transformed bythe natural logarithm.

Fig. 4 Cumulative distribution curves of PM10 and PM2.5 in-door levels in the summer and winter monitoring sessions.Twenty-four-hour Air Quality guidelines of World HealthOrganization for PM10 and PM2.5 (solid lines) are reported ascomparison references

Airborne particulate matter and gaseous air pollutants

495

set of measurements for NO2 showed an indoor increaseof 4.5% per 10% increase in outdoor concentration.A CO regression model showed a non-significant

relationship for summer monitoring and importantindoor variations (on average 11.7%) related tooutdoor increases of 10% for winter monitoring, whichdecreased to 8.0% when the two monitoring sessionswere considered together.The spatial distribution of PM2.5 and PM10 in

outdoor environments is generally homogeneous(Monn, 2001), as confirmed by the good correspon-dence between residential outdoor and ambient levels,with Pearson�s coefficients (r) >0.69 in the worst caseand generally >0.9 (Brown et al., 2008; Evans et al.,2000; Oglesby et al., 2000; Rodes et al., 2010; Williamset al., 2000). On the contrary, gaseous pollutants aresubject to considerable small-scale spatial variabilitywithin urban areas (Monn, 2001). Thus, ambient datawere used as proxies for outdoor PM concentrations inthis study, and indoor/outdoor (I/O) ratios were calcu-lated for each monitored dwelling. Mean (standarddeviation) estimated I/O ratios for PM2.5 and PM10

were 1.22 (0.91) and 1.04 (0.76), respectively, for thestudy as a whole, 1.14 (0.93) and 1.13 (0.73) in summerand 1.30 (0.90) and 0.93 (0.78) in winter, respectively.In summer, FINF was 0.47 for PM2.5 and 0.41 for

PM10, with Cig of 12.5 lg/m3 for PM2.5 and 17.9 lg/m3

for PM10. In winter, an increase in FINF and Cig wasestimated with respect to summer. This is particularlytrue for PM2.5, whose FINF and Cig increased by 61%and 30%, respectively, against a 15% and 26%increase in PM10. The estimated seasonal PM2.5 andPM10 contributions of non-ambient (Cig) and ambient(Cog) origins to the respective total PM mass concen-trations are depicted in Figure 6.

Discussion

PM sampler comparison

GK cyclones and PCIS were characterized by recipro-cal predictability as regards the inter-comparisoncriteria set forth by Watson et al. (1998). However,indoor concentrations measured by GK samplers(especially GK2.05) were on average considerablylower than those simultaneously measured by PCIS(Figure 3). The disagreement between GK and PCISsamplers could be due to possible deterioration ordeformation of GK plastic cassette filter holders, whichwere periodically cleaned but not replaced with newones during the whole study. This could lead to airleaks in GK systems with a consequent decrease insampling efficiency.

Comparison of airborne pollutant concentrations with historical data

On average, measured indoor PM2.5 levels (Table 2)were lower ()22%) compared with those measuredabout 10 years earlier (33.2 vs. 42.7 lg/m3) in buildingsin Milan (approximately 40 km from Lodi) during theEXPOLIS study (Rotko et al., 2002). The Milanmetropolitan area is possibly the best comparisonscenario for PM, owing to its geographic proximity toLodi and its very similar meteorological conditions.However, emission sources of PM typical of Lodiprovince are likely to be somewhat different than thoseof Milan, because Lodi lies in a rural area mainlycharacterized by agriculture and animal husbandry;automotive traffic exhausts and civil heating emissionsrepresent the main sources of PM pollution in Milan.The decrease in indoor PM2.5 concentrations over timeis in accordance with lower outdoor concentrations(23.0 vs. 41.3 lg/m3) (Rotko et al., 2002) and the slightdecreasing trend in PM10 ambient levels in Lombardyfrom 2002 to 2008 [Environmental Protection Agencyof Lombardy (ARPA), 2008]. Indoor PM2.5 meanlevels reported in Table 2 are generally higher thanthose found in northern Europe and the United States(Adgate et al., 2003; Brown et al., 2008; Evans et al.,2000; Hanninen et al., 2004; Janssen et al., 2000; Laiet al., 2004; Meng et al., 2005; Rodes et al., 2010;Rojas-Bracho et al., 2000; Williams et al., 2000); this ismainly because of the peculiar demographic andmeteorological conditions of the Po plain, which ischaracterized by high population density and highatmospheric stability (low mixing layer height and lowwind speed), which lead to accumulation of outdooremitted pollutants (Vecchi et al., 2004). However,mean PM2.5 indoor concentrations measured in Lodiprovince were quite comparable with those measuredin central and southern Europe (Athens and Prague)during the 1996–2000 period (Hanninen et al., 2004). Itshould be noted that, as stated above, indoor atmo-spheric pollutant concentrations may have changed in

Fig. 6 Estimated seasonal contributions of Cig and Cog to in-door measured concentrations of PM2.5 and PM10. Absolute(microgram/cubic meter) and relative values (%) are reported inthe histograms

Cattaneo et al.

496

the last decade. This could make the comparison withhistorical data somewhat biased, depending on the typeof pollutant.Mean indoor NO2 concentrations measured in living

rooms in the Lodi area (Table 2) are within the rangeof those measured in bedrooms (25 lg/m3) and kitch-ens (47 lg/m3) in Genoa (Gallelli et al., 2002), about150 km from Lodi. On average, indoor levels of NO2

were comparable or slightly higher than those histor-ically reported in northern and central Europe (Laiet al., 2004), with the exception of Prague (Kousaet al., 2001). Moreover, median indoor NO2 levelsmeasured in the present work were 7–11 lg/m3 lowerthan those reported in a 2003–2006 Spanish study.The range of indoor O3 levels monitored in this study

was comparable with those (0–18 ppb) measured inHelsinki in 1999 (Koponen et al., 2001) and in U.S.indoor environments (0–41 lg/m3) in spring 1996 (Leeet al., 2002).Average levels of indoor CO in Lodi province were

lower than those measured about 10 years ago inMilan in the EXPOLIS project (1.9 ppm) (BruinenDe Bruin et al., 2004), which is explainable by thedecreasing trend of outdoor concentrations in Lomb-ardy over the last decade [Environmental ProtectionAgency of Lombardy (ARPA), 2008], thanks to theeffect of the adoption of catalytic converters in vehicleexhaust systems.

Comparison of indoor concentrations with GV

WHO AQG 24-h values for PM2.5 (Figure 4) wereexceeded in 46% of total cases (72% in winter and 22%in summer), which may imply an excess of morbidityand mortality for the population. Moreover, the PM2.5

interim target 3 (35 lg/m3) was not achieved in 29% ofcases, implying an estimated 1.2% increase in short-term mortality [World Health Organization (WHO),2006]. At present, no specific GV are developed for riskassessment and management of PM in indoor environ-ments, even though different patterns of health effectsbetween particles of ambient and non-ambient originshave been observed (Ebelt et al., 2005; Koenig et al.,2005); this should be taken into account in futureregulations and research. PM10 levels in indoor envi-ronments in Lodi province were not compared withPM10 GV, as PM10 is intrinsically less relevant thanPM2.5 in terms of health effects on humans. As a matterof fact, the PM10 GV recommended by WHO weremerely developed via application of a PM2.5/PM10

ratio of 0.5 [World Health Organization (WHO),2006].It is reasonable to expect NO2 annual WHO GV

(40 lg/m3) to be exceeded in a significant proportion(about 20%) of studied dwellings. Long-term expo-sures to these low levels may involve direct toxic effects,even if it is still unclear to what extent the health effects

observed in epidemiological studies are attributable toNO2 itself or to other combustion-related products[World Health Organization (WHO), 2006].The CO2 ASHRAE GV was exceeded with low

frequency (12% of cases), especially in winter. Thisthreshold value was proposed on the basis of theassociation of CO2 concentrations with undesirablelevels of body odor. Thus, this value is not based onany health impact of carbon dioxide itself on humans[American Society of Heating Refrigerating and Air-Conditioning Engineers (ASHRAE), 1989].Carbon monoxide daily and short-term (8 h, 30, and

15 min) averages and O3 concentrations never ex-ceeded the recommended GV in the monitored homes.

Seasonal pollutant trends

Indoor winter concentrations of all the consideredpollutants were higher than corresponding summerlevels (with the exception of O3), which is consistentwith ambient seasonal variations (Table 2) related tohigher local emission rates and higher atmosphericstability in winter (Vecchi et al., 2004). In the case ofPM, this is reinforced by the fact that estimated Cog forPM2.5 and PM10 indoors in winter were significantlyhigher (about 3-fold) than corresponding Cog insummer (Figure 6). Nevertheless, lower AER (Table 2)and higher indoor source strengths (heating systems)could have played a role in contributing to the increasein indoor PM levels in winter with respect to summer,as confirmed by the slight growth (25–30%) inestimated Cig (Figure 6).

Size-fractionated PM

Among the indoor PM fractions, PM0.5 is predominantin terms of mass concentrations (Figure 5). Suchparticles are assumed to derive mainly from conden-sation and accumulation mechanisms outdoors (Will-eke and Whitby, 1975). The 600 000 head of cattleraised in Lodi province may contribute to highemissions of ammonia. Consequently, compoundsresulting from reactions of ammonia with other gasessuch as sulfuric and nitric acids may strongly contrib-ute to PM of secondary origin; this is assumed to be animportant component of total PM (Derwent andHertel, 1998). On the other hand, indoor sources canalso represent an important contribution to indoorPM0.5 concentrations, as they generally generate fineand ultrafine particles (Afshari et al., 2005). Themeasured PM0.5 concentrations in sample homes areeven more dominant in winter (Figure 5); this isprobably due to an additional quantity of fine particlesinfiltrated from outdoors along with those generatedby season-specific indoor combustion sources, such asboilers and other heating equipment. Such findings arein agreement with the seasonal contributions of PM0.5

Airborne particulate matter and gaseous air pollutants

497

to the individual concentrations of PM10 measured inthe Milan metropolitan area (54% in summer and 68%in winter), as part of the PM-CARE project (Schlittet al., 2008). The lower amount of airborne coarseparticles in winter may be related to tighter sealing ofhomes (see AER in Table 2), which selectively preventsinfiltration of coarse particles from outdoor air, as thepenetration of outdoor particles indoors generallydecreases with size (Abt et al., 2000).

Influence of ambient levels on indoor concentrations

Ambient data explained the variability of indoor levelsin Lodi province (Table 5) in a manner consistent withthe results of the RIOPA study, in which R2 between0.06 and 0.44 were found in three US cities (Menget al., 2005). On the contrary, the explained variabilityof PM2.5 indoor concentrations by ambient levels wasfound to be lower than that reported for four Euro-pean cities (Hanninen et al., 2004), in a study restrictedto non-smoking homes. In contrast, we examined asignificant number of residential dwellings with possi-ble contamination by environmental tobacco smoke(Table 1). Other research shows that outdoor sourcescan explain up to 65% of indoor concentrations(Brown et al., 2008). In this case, the study designwas based on many measurements per home, and datapoints characterized by extreme cooking events andhumidifier use were excluded. Both of these factorscould lead to an improvement in indoor-ambientcorrelations. Possible other reasons that explain differ-ences in the percentage of explained variability ofindoor levels by outdoor concentrations include thediffering impact of source strength and differinginfiltration factors from place to place. Moreover, itshould be remembered that in the present studyambient levels were adopted as proxies of outdoorconcentrations, and small differences in mass concen-trations between outdoor and ambient PM levelsshould be expected, as the spatial distribution of fineparticles is relatively uniform (Monn, 2001).The explained variability and the strength of asso-

ciation between PM2.5 indoor and outdoor levels arebetter in winter with respect to summer, consistent withthe findings of Brown et al. (2008). They interpretedthis finding as an indication that indoor PM2.5 sourceshave a greater impact on PM2.5 exposures duringwinter when homes have reduced ventilation.In both seasons, a similar absolute contribution of

Cog to indoor PM2.5 and PM10 was found (Figure 6).This trend may be explained by the low penetrationrate of coarse particles, owing to their tendency tosettle onto upward facing surfaces (Nazaroff, 2004). Onthe contrary, PM10 Cig was measurably higher thanthat of PM2.5 (43% increase in summer and 39% inwinter), hence highlighting the major role of coarseparticles generated indoors with respect to fine particles

(Figure 6) in determining corresponding indoor con-centrations, in agreement with the findings of Abt et al.(2000).The correlation between indoor and ambient levels

of gaseous pollutants showed worse regression anddetermination (r2) coefficients than other publisheddata, with the exception of the CO regression coeffi-cient in winter. This is primarily owing to the fact thatambient levels of gaseous compounds are not goodproxies for outdoor levels, which are characterized byhigh spatial variability (Monn, 2001).Mean and median PM2.5 I/O ratios were greater in

winter than in summer, while PM10 I/O ratios showedan opposite trend (Table 2). In the case of PM2.5, suchwinter I/O increase was interpreted to be related toadditional seasonal sources, such as domestic heatingsystems located indoors and wood burning, which mayalso explain the estimated concentrations of indoor-generated PM2.5 (Figure 6) and the high proportion ofcases (51%) with I/O > 1. Accumulation mode parti-cles should be originated mainly outdoors especially inwinter, in accordance with Cog reported in Figure 6. Asstated before, PM10 I/O seasonal trends are opposite tothose of PM2.5 and may be reasonably explained by theaforementioned coarse fraction behavior, which ischaracterized by higher indoor levels in summer(Figure 5).

Conclusion

Particulate matter and gaseous pollutants were mea-sured twice in 53 residential structures in the provinceof Lodi, in the Po Valley of Italy, which is character-ized by high pollution levels. PM and NO2 levelsmeasured in residential dwellings were in some caseshigher than GV recommended by the World HealthOrganization (WHO) (2006). On the contrary, CO andO3 indoor concentrations were always below therecommended thresholds.All the study pollutants (except for coarse PM and

O3) showed a marked seasonal trend, with higherwinter levels in comparison with corresponding sum-mer concentrations, in agreement with simultaneousambient measurements. In the case of PM, seasonaltrends were related to the substantial increase inconcentrations of PM0.5, which represents particlesgenerated by nucleation and accumulation-coagula-tion processes. Indoor PM seasonal differences weremainly linked to the contribution of particles ofambient origin, accounting for 63% of PM2.5 and58% of PM10 in winter. On the contrary, in summer,indoor PM levels comprised mainly particles generatedindoors. Moreover, indoor concentrations of coarseparticles were interpreted to be mainly of indoororigin, as Cog for PM2.5 and PM10 were similar, whilecorresponding Cig showed a substantial increase fromPM2.5 to PM10.

Cattaneo et al.

498

Future research should investigate chemical constit-uents of fine PM, with a particular focus on secondarysemi-volatile compounds linked to NH3 atmosphericemissions, which are very high in the province of Lodi.Our study of the most relevant structural and behav-ioral determinants of indoor concentrations and thequantification of their relative contributions to indoorpollutant levels will be presented in a companion paper.Results from such studies could prove useful inprogramming proper risk management and communi-cation interventions by local health protection agencies.

Acknowledgements

This research was funded by the Local Health Unit ofthe province of Lodi. We thank Eugenio Ariano forproject coordination, Salvatore Pulvirenti, FrancoVercelli, and all the technical staff at the Local HealthUnit of the province of Lodi for their critical help insample collection. Finally, we thank Dario Cantoniand Marco Gurini for the NO2 and O3 analyses.

References

Abt, E., Suh, H.H., Allen, G. and Koutrakis,P. (2000) Characterization of indoorparticle sources: a study conducted in themetropolitan Boston area, Environ.Health Perspect., 108, 35–44.

Adgate, J.L., Ramachandran, G., Pratt,G.C., Waller, L.A. and Sexton, K. (2003)Longitudinal variability in outdoor,indoor, and personal PM2.5 exposurein healthy non-smoking adults, Atmos.Environ., 37, 993–1002.

Afshari, A., Matson, U. and Ekberg, L.E.(2005) Characterization of indoor sourcesof fine and ultrafine particles: a studyconducted in a full-scale chamber, IndoorAir, 15, 141–150.

Altman, D. and Bland, J. (1983) Measure-ment in medicine: the analysis of methodcomparison studies, Statistician, 32, 307–317.

American Society of Heating Refrigeratingand Air-Conditioning Engineers (ASH-RAE) (1989) Ventilation for AcceptableIndoor Air Quality, Atlanta, GA, USA,ASHRAE, 33.

American Society of Testing and Materials(ASTM) (2000) ASTM Standard Practicefor Controlling and Characterizing Errorsin Weighing Collected Aerosols (ASTM D6552), West Conshohocken, PA, USA,ASTM.

Bland, J.M. and Altman, D.G. (1986) Sta-tistical methods for assessing agreementbetween two methods of clinical mea-surement, Lancet, 1, 307–310.

Brown, K.W., Sarnat, J.A., Suh, H.H.,Coull, B.A., Spengler, J.D. and Koutra-kis, P. (2008) Ambient site, home outdoorand home indoor particulate concentra-tions as proxies of personal exposures,J. Environ. Monit., 10, 1041–1051.

Brown, K.W., Sarnat, J.A., Suh, H.H.,Coull, B.A. and Koutrakis, P. (2009)Factors influencing relationships betweenpersonal and ambient concentrations ofgaseous and particulate pollutants, Sci.Total Environ., 407, 3754–3765.

Bruinen De Bruin, Y., Hanninen, O., Carrer,P., Maroni, M., Kephalopoulos, S.,Scotto Di Marco, G. and Jantunen, M.

(2004) Simulation of working populationexposures to carbon monoxide usingEXPOLIS-Milan microenvironment con-centration and time-activity data, J. Expo.Anal. Environ. Epidemiol., 14, 154–163.

Cocheo, V., Boaretto, C. and Sacco, P.(1996) High uptake rate radial diffusivesampler suitable for both solvent andthermal desorption, Am. Ind. Hyg. Assoc.J., 57, 897–904.

Delfino, R.J., Sioutas, C. and Malik, S.(2005) Potential role of ultrafine particlesin associations between airborne particlemass and cardiovascular health, Environ.Health Perspect., 113, 934–946.

Derwent, R. and Hertel, O. (1998) Trans-formation of air pollutants, In: Fenger, J.,Hertel, O. and Palmgren, S. (eds) UrbanAir Pollution-European Aspects, Dordr-echt, Netherlands, Kluwer AcademicPublishers, 137–160.

Dockery, D.W., Pope, C.A., Xu, X., Spen-gler, J.D., Ware, J.H., Fay, M.E., Ferris,B.G. and Speizer, F.E. (1993) An associ-ation between air pollution and mortalityin six U.S. cities, N. Engl. J. Med., 329,1753–1759.

Ebelt, S.T., Wilson, W.E. and Brauer, M.(2005) Exposure to ambient and nonam-bient components of particulate matter: acomparison of health effects, Epidemiol-ogy, 16, 396–405.

Environmental Protection Agency of Lomb-ardy (ARPA) (2008) Rapporto sullaqualita dell�aria di Milano e provincia –Anno 2008, Available on-line at http://ita.arpalombardia.it/ita/qaria/pdf/RQA-2008/RQA_MI_2008.pdf, accessed onApril, 2011.

Evans, G.F., Highsmith, R.V., Sheldon, L.S.,Suggs, J.C., Williams, R.W., Zweidinger,R.B., Creason, J.P., Walsh, D., Rodes,C.E. and Lawless, P.A. (2000) The 1999Fresno particulate matter exposure stud-ies: comparison of community, outdoor,and residential PM mass measurements,J. Air Waste Manag. Assoc., 50, 1887–1896.

Gallelli, G., Orlando, P., Perdelli, F. andPanatto, D. (2002) Factors affecting

individual exposure to NO2 in Genoa(northern Italy), Sci. Total Environ., 287,31–36.

Georgoulis, L.B., Hanninen, O., Samoli, E.,Katsouyanni, K., Kunzli, N., Polanska,L., Bruinen de Bruin, Y., Alm, S. andJantunen, M. (2002) Personal carbonmonoxide exposure in five European cit-ies and its determinants, Atmos. Environ.,36, 963–974.

Hanninen, O.O., Lebret, E., Ilacqua, V.,Katsouyanni, K., Kunzli, N., Sram, R.J.and Jantunen, M. (2004) Infiltration ofambient PM2.5 and levels of indoor gen-erated non-ETS PM2.5 in residences offour European cities, Atmos. Environ., 38,6411–6423.

Janssen, N.A.H., De Hartog, J.J., Hoek, G.,Brunekreef, B., Lanki, T., Timonen, K.L.and Pekkanen, J. (2000) Personal expo-sure to fine particulate matter in elderlysubjects: relation between personal,indoor, and outdoor concentrations, J. AirWaste Manag. Assoc., 50, 1133–1143.

Janssen, N.A.H., Lanki, T., Hoek, G., Val-lius, M., De Hartog, J.J., Van Grieken,R., Pekkanen, J. and Brunekreef, B.(2005) Associations between ambient,personal, and indoor exposure to fineparticulate matter constituents in Dutchand Finnish panels of cardiovascularpatients, Occup. Environ. Med., 62, 868–877.

Koenig, J.Q., Mar, T.F., Allen, R.W.,Jansen, K., Lumley, T., Sullivan, J.H.,Trenga, C.A., Larson, T. and Liu, L.J.(2005) Pulmonary effects of indoor- andoutdoor-generated particles in childrenwith asthma, Environ. Health Perspect.,113, 499–503.

Koistinen, K.J., Kousa, A., Tenhola, V.,Hanninen, O., Jantunen, M.J., Oglesby,L., Kuenzli, N. and Georgoulis, L. (1999)Fine particle (PM2.5) measurement meth-odology, quality assurance procedures,and pilot results of the EXPOLIS study,J. Air Waste Manag. Assoc., 49, 1212–1220.

Koponen, I.K., Asmi, A., Keronen, P.,Puhto, K. and Kulmala, M. (2001)

Airborne particulate matter and gaseous air pollutants

499

Indoor air measurement campaign inHelsinki, Finland 1999 – the effectof outdoor air pollution on indoorair, Atmos. Environ., 35, 1465–1477.

Kousa, A., Monn, C., Rotko, T., Alm, S.,Oglesby, L. and Jantunen, M.J. (2001)Personal exposures to NO2 in theEXPOLIS-study: relation to residentialindoor, outdoor and workplaceconcentrations in Basel, Helsinki andPrague, Atmos. Environ., 35, 3405–3412.

Lai, H.K., Kendall, M., Ferrier, H., Lindup,I., Alm, S., Hanninen, O., Jantunen, M.,Mathys, P., Colvile, R., Ashmore, M.R.,Cullinan, P. and Nieuwenhuijsen, M.J.(2004) Personal exposures and microen-vironment concentrations of PM2.5, VOC,NO2 and CO in Oxford, UK, Atmos.Environ., 38, 6399–6410.

Lee, K., Xue, J., Geyh, A.S., Ozkaynak, H.,Leaderer, B.P., Weschler, C.J. and Spen-gler, J.D. (2002) Nitrous acid, nitrogendioxide, and ozone concentrations in res-idential environments, Environ. HealthPerspect., 110, 145–150.

Meng, Q.Y., Turpin, B.J., Korn, L., Weisel,C.P., Morandi, M., Colome, S., Zhang,J., Stock, T., Spektor, D., Winer, A.,Zhang, L., Lee, J.H., Giovanetti, R., Cui,W., Kwon, J., Alimokhtari, S., Shendell,D., Jones, J., Farrar, C. and Maberti, S.(2005) Influence of ambient (outdoor)sources on residential indoor and per-sonal PM2.5 concentrations. Analyses ofRIOPA data, J. Expo. Anal. Environ.Epidemiol., 15, 17–28.

Misra, C., Singh, M., Shen, S., Sioutas, C.and Hall, P.M. (2002) Development andevaluation of a personal cascade impactorsampler (PCIS), J. Aerosol Sci., 33, 1027–1047.

Monn, C. (2001) Exposure assessment of airpollutants: a review on spatial heteroge-neity and indoor/outdoor/personal expo-sure to suspended particulate matter,nitrogen dioxide and ozone, Atmos.Environ., 35, 1–32.

Nazaroff, W.W. (2004) Indoor particledynamics, Indoor Air, 14(Suppl 7), 175–183.

Oglesby, L., Kunzli, N., Roosli, M., Braun-Fahrlander, C., Mathys, P., Stern, W.,Jantunen, M. and Kousa, A. (2000)Validity of ambient levels of fine particlesas surrogate for personal exposure tooutdoor air pollution. Results of theEuropean EXPOLIS-EAS study (Swiss

Center Basel), J. Air Waste Manag.Assoc., 50, 1251–1261.

Osman, L.M., Douglas, J.G., Garden, C.,Reglitz, K., Lyon, J., Gordon, S. andAyres, J.G. (2007) Indoor air quality inhomes of patients with chronic obstruc-tive pulmonary disease, Am. J. Resp. Crit.Care Med., 176, 465–472.

Ott, W., Wallace, L. and Mage, D. (2000)Predicting particulate (PM10) personalexposure distributions using a randomcomponent superposition statisticalmodel, J. Air Waste Manage Assoc., 50,1398–1406.

Polidori, A., Arhami, M., Sioutas, C.,Delfino, R.J. and Allen, R. (2007) Indoor/Outdoor relationships, trends, andcarbonaceous content of fine particulatematter in retirement homes of the LosAngeles Basin, J. Air Waste Manag.Assoc., 57, 366–379.

Pope, C.A. III (2000) Review: epidemiologi-cal basis for particulate air pollutionhealth standards, Aerosol Sci. Technol.,32, 4–14.

Rodes, C.E., Lawless, P.A., Thornburg, J.W.,Williams, R.W. and Croghan, C.W. (2010)DEARS particulate matter relationshipsfor personal, indoor, outdoor, and centralsite settings for a general population,Atmos. Environ., 44, 1386–1399.

Rojas-Bracho, L., Suh, H.H. and Koutrakis,P. (2000) Relationships among personal,indoor, and outdoor fine and coarseparticle concentrations for individualswith COPD, J. Expo. Anal. Environ.Epidemiol., 10, 294–306.

Rotko, T., Oglesby, L., Kunzli, N., Carrer,P., Nieuwenhuijsen, M.J. and Jantunen,M. (2002) Determinants of perceived airpollution annoyance and associationbetween annoyance scores and airpollution (PM2.5, NO2) concentrations inthe European EXPOLIS study, Atmos.Environ., 36, 4593–4602.

Sarnat, J.A., Koutrakis, P. and Suh, H.H.(2000) Assessing the relationship betweenpersonal particulate and gaseous expo-sures of senior citizens living in Baltimore,MD, J. Air Waste Manag. Assoc., 50,1184–1198.

Schlitt, C., Garramone, G., Cattaneo, A.,Peruzzo, C., Taronna, M., Pulvirenti, S.,Vercelli, F. and Cavallo, D. (2008) Fineand ultrafine particle levels determinedduring everyday activities: the PM-CAREproject. In: Strøm-Tejsen, P., Olesen,B.W., Wargocki, P., Zukowska, D. andToftum, J. (eds) Proceedings of Indoor Air

’08, 11th International Conference onIndoor Air Quality and Climate,Copenhagen, Denmark.

Singh, M., Misra, C. and Sioutas, C. (2003)Field evaluation of a personal cascadeimpactor sampler (PCIS), Atmos. Envi-ron., 37, 4781–4793.

Sioutas, C., Delfino, R.J. and Singh, M.(2005) Exposure assessment for atmo-spheric ultrafine particles (UFPs)and implications in epidemiologicresearch, Environ. Health Perspect., 113,947–955.

Vecchi, R., Marcazzan, G., Valli, G.,Ceriani, M. and Antoniazzi, C. (2004)The role of atmospheric dispersion in theseasonal variation of PM1 and PM2.5

concentration and composition in theurban area of Milan (Italy), Atmos.Environ., 38, 4437–4446.

Watson, J.G., Chow, J.C., Moosmuller, H.,Green, M., Frank, N. and Pitchford, M.(1998) Guidance for using continu-ous monitors in PM2.5 monitoringnetworks, U.S. Environ Prot Agency, OffAir Qual Plann Stand, [Tech. Rep.] EPA,ii–xv, 1.

Weschler, C.J. (2006) Ozone�s impact onpublic health: contributions from indoorexposures to ozone and products ofozone-initiated chemistry, Environ.Health Perspect., 114, 1489–1496.

WHO (2010) WHO Guidelines for Indoor AirQuality: Selected Pollutants, Copenhagen,Denmark, WHO Regional Office forEurope.

Willeke, K. and Whitby, K. (1975)Atmospheric aerosols: size distributioninterpretation, J. Air Pollut. ControlAssoc., 25, 529–534.

Williams, R., Creason, J., Zweidinger, R.,Watts, R., Sheldon, L. and Shy, C. (2000)Indoor, outdoor, and personal exposuremonitoring of particulate air pollution:the Baltimore elderly epidemiology-expo-sure pilot study, Atmos. Environ., 34,4193–4204.

World Health Organization (WHO) (2006)WHO Air Quality Guidelines for Particu-late Matter, Ozone, Nitrogen Dioxide andSulfur Dioxide: Global Update 2005(Summary of Risk Assessment), Availableonline at http://whqlibdoc.who.int/hq/2006/WHO_SDE_PHE_OE-H_06.02_eng.pdf, accessed on April 2011.

Zou, B., Wilson, J.G., Zhan, F.B. and Zeng,Y. (2009) Air pollution exposure assess-ment methods utilized in epidemiologicalstudies, J. Environ. Monit., 11, 475–490.

Cattaneo et al.

500