Cost-Effectiveness of Air Purifiers against Pollution

The goal for this post is to give an in­tro­duc­tion into the hu­man health effects of air pol­lu­tion, en­courage fur­ther dis­cus­sion, and eval­u­ate an in­ter­ven­tion: The use of air puri­fiers in homes. Th­ese air puri­fiers are in­ex­pen­sive, stan­dalone de­vices not re­quiring any spe­cial in­stal­la­tion pro­ce­dure.

This par­tic­u­lar in­ter­ven­tion was se­lected out of per­sonal in­ter­est, not be­cause I be­lieve it’s par­tic­u­larly effec­tive. It’s plau­si­ble that other in­ter­ven­tions against air pol­lu­tion would be much bet­ter—for ex­am­ple, pro­vid­ing more peo­ple with clean en­ergy for cook­ing.

We will in­ves­ti­gate what it costs to sig­nifi­cantly re­duce per­sonal ex­po­sure to the most dam­ag­ing form of par­tic­u­late mat­ter (PM2.5) us­ing these de­vices. A first anal­y­sis sug­gests that the cost-effec­tive­ness of this in­ter­ven­tion is two or­ders of mag­ni­tude worse than the best EA in­ter­ven­tions. How­ever, it is still good enough to qual­ify as an “effec­tive” or even “highly effec­tive” health in­ter­ven­tion ac­cord­ing to WHO crite­ria.

The model cor­re­spond­ing to this post can be found here: https://​​www.getguessti­mate.com/​​mod­els/​​16649

Epistemic sta­tus: I am con­fi­dent that this post iden­ti­fies most of the big ques­tions and un­cer­tain­ties. Given that I have no back­ground in pub­lic health, it is pos­si­ble I’ve missed ma­jor pieces of the puz­zle, and it’s likely that the spe­cific num­bers are off.

Quan­tify­ing hu­man health im­pacts of air pollution

Air pol­lu­tion is a sig­nifi­cant risk fac­tor for car­dio­vas­cu­lar dis­ease, can­cer, res­pi­ra­tory in­fec­tions and COPD (WHO). For fur­ther dis­cus­sion, we will look at par­tic­u­late mat­ter pol­lu­tion (PM) only—which, by it­self, caused a loss of 106 mil­lion DALYs (dis­abil­ity ad­justed life years) in 2016 − 51% of that in China and In­dia (State of Global Air Re­port 2018). A Fin­nish study looked at NO2 and O3 in ad­di­tion to PM, and found that PM con­tributes 85% of the to­tal dis­ease bur­den among those pol­lu­tants, with the finest par­ti­cles (PM2.5, par­ti­cles with a di­ame­ter of less than 2.5µm) pro­duc­ing the vast ma­jor­ity of the harm. There­fore, for the re­main­der of this post we will only con­sider PM2.5, which is mea­sured in µg/​m³.

Air pol­lu­tion is a sig­nifi­cant prob­lem even in re­gions with rel­a­tively good air: The WHO states that “even in the Euro­pean Union, where PM con­cen­tra­tions in many cities do com­ply with guideline lev­els, it is es­ti­mated that av­er­age life ex­pec­tancy is 8.6 months lower than it would oth­er­wise be, due to PM ex­po­sures from hu­man sources.”

The role of ul­tra­fine particles

Ac­cord­ing to some sources, ul­tra­fine par­ti­cles—which are sig­nifi­cantly smaller than 2.5µm—might pro­duce a sig­nifi­cant por­tion of the harm. How­ever, for our in­ves­ti­ga­tion here it is enough to con­sider PM2.5 mea­sure­ments only, with­out gath­er­ing sep­a­rate data on ul­tra­fine par­ti­cles. This is for two rea­sons: First, we can as­sume that all kinds of PM are roughly cor­re­lated with each other (most liter­a­ture uses PM2.5 mea­sure­ments only). Se­cond, the HEPA filter in­ter­ven­tion we will sug­gest is good at fil­ter­ing ul­tra­fine par­ti­cles—so if PM2.5 is suc­cess­fully filtered, ul­tra­fine par­ti­cles are re­moved too. [1]

Ex­ist­ing work

A large amount of stud­ies in­ves­ti­gate the re­la­tion­ship be­tween PM2.5 ex­po­sure lev­els and health effects. The con­sen­sus seems to be that this re­la­tion­ship is sub­lin­ear when look­ing at a large range: at high con­cen­tra­tions, each ad­di­tional µg/​m³ of PM2.5 con­tributes less harm. Here are some stud­ies which ex­am­ine this non­lin­ear re­la­tion­ship: 1, 2, 3, 4, 5, 6, 7.

All of those stud­ies calcu­late ex­tra mor­tal­ity or rate of ill­ness. This data is not suffi­cient to es­ti­mate years of life lost or DALYs [2]. Other stud­ies do make es­ti­mates of these, but they do not take into ac­count the non­lin­ear­ity:

  • In the EU: an ad­di­tional 1µg/​m³ of PM2.5 for one year, lead­ing to a loss of life ex­pec­tancy of 0.22 days per per­son in the EU.

  • In In­dia: an av­er­age level of of 89.9 µg/​m³, lead­ing to a loss of 1.6 years of life per per­son. [3]

  • In the UK: A loss of life of 2.7 days per per­son at a level of 9.9µg/​m³ for one year (this agrees very closely with the EU val­ues men­tioned above).

In­door vs Out­door Pol­lu­tion Levels

In­door air pol­lu­tion cor­re­lates with out­door air pol­lu­tion rel­a­tively closely, rang­ing from “50% lower” to “as high if not higher”, de­pend­ing on hu­man ac­tivity (source). Peo­ple us­ing open stoves for cook­ing will ex­pe­rience the high­est lev­els of air pol­lu­tion—far above the amount of PM2.5 even in bad city air (source). Re­searchers in Ger­many mea­sured a ra­tio be­tween in­door and out­door PM2.5 con­cen­tra­tion of 0.33-0.78 - the lower value for closed win­dows, the lat­ter value for tilted win­dows. How­ever those mea­sure­ments were only performed in an un­in­hab­ited room in one build­ing in one city.

The spe­cific ra­tio be­tween in­door and out­door PM2.5 con­cen­tra­tion might de­pend on a va­ri­ety of fac­tors:

  • Amount of ven­tila­tion through opened windows

  • Amount of air in­fil­tra­tion when win­dows are closed

  • Ab­solute level of out­side pollution

  • In­door pol­lu­tion sources (cook­ing, smok­ing, etc…)

  • Num­ber of oc­cu­pants and pets

A com­pre­hen­sive meta study looked at 61 ar­ti­cles in­ves­ti­gat­ing differ­ent kinds of build­ings. Over 40% of ar­ti­cles found higher in­doors than out­doors PM2.5 lev­els. The au­thors of the meta study do not at­tempt to ar­rive at a con­sen­sus for the ra­tio be­tween out­door and in­door lev­els, but we can guess that it’s plau­si­ble to as­sume in­door lev­els equal­ling out­door lev­els (from Figure 3a in the meta study). For smok­ing house­holds, this as­sump­tion is likely to be too low, for non-smok­ing house­holds with low air in­fil­tra­tion from the out­side this might be too high.

For our calcu­la­tion we don’t even need to con­sider in­door lev­els ex­plic­itely, for rea­sons de­scribed in the next sec­tion.

Per­sonal Ex­po­sure Levels

The rele­vant quan­tity for health effects is the per­sonal ex­po­sure a per­son ex­pe­riences dur­ing a cer­tain pe­riod of time. We will not con­sider this quan­tity di­rectly. In­stead, we will es­ti­mate the health effects for cer­tain out­door PM2.5 lev­els us­ing real-world data. This data already in­cludes the fact that peo­ple spend their days in a va­ri­ety of in­door and out­door set­tings. There­fore, the re­la­tion­ship be­tween out­door and per­sonal ex­po­sure is already im­plic­itly taken into ac­count.

The same can be said for in­door lev­els, so we don’t need to es­ti­mate those ei­ther.

Difficul­ties in es­ti­mat­ing the effects of PM2.5 reduction

We will pro­pose to place air puri­fiers in sub­jects homes, which gives them sig­nifi­cantly re­duced PM2.5 lev­els for parts of the day. How can we es­ti­mate the health benefits of this in­ter­ven­tion? There are two ob­sta­cles here.

Ob­sta­cle 1: Mea­sure­ment of mor­tal­ity vs years of life lost

We know that health im­pact scales sub­lin­early with PM2.5 lev­els. In our sum­mary of ex­ist­ing work, we have seen that most stud­ies that eval­u­ate this non­lin­ear re­la­tion­ship es­ti­mate the mor­tal­ity risk only and do not at­tempt to quan­tify the years of life lost. This is an ob­sta­cle to our anal­y­sis. In or­der to ar­rive at cost-effec­tive­ness es­ti­mates, we would very much like to know the ad­di­tional years of healthy life that can be gen­er­ated per $ in­vested in our in­ter­ven­tion. The stud­ies eval­u­ated do not provide enough in­for­ma­tion to calcu­late that. None of the stud­ies con­sid­ered calcu­lates the gain in years of life at each PM2.5 level while cor­rectly tak­ing into ac­count the sub­lin­ear na­ture of the re­la­tion­ship.

This could be re­solved in sev­eral ways:

  1. Do a proper anal­y­sis, in­cor­po­rat­ing ad­di­tional data. [4]

  2. Take re­li­able DALY val­ues from one source and mul­ti­ply it with the shape of the mor­tal­ity risk curve taken from other sources.

  3. Use a sim­ple rule of thumb: For ex­am­ple, this UK study sug­gests mul­ti­ply­ing the num­ber of pre­ma­ture deaths by 12 to ar­rive at the years of life lost (of course, this rule-of-thumb is valid only in the par­tic­u­lar case of air pol­lu­tion).

For our in­ves­ti­ga­tion here, we choose nei­ther of these ap­proaches and sim­ply ig­nore the non­lin­ear­ity for rea­sons that will be out­lined be­low.

Ob­sta­cle 2: Is av­er­age PM2.5 level the right kind of anal­y­sis?

Even if ob­sta­cle 1 did not ap­ply and we knew the re­la­tion­ship be­tween av­er­age PM2.5 and years of life lost perfectly, this data would not be di­rectly ap­pli­ca­ble to our anal­y­sis. This is be­cause we are not re­duc­ing av­er­age PM2.5: We are sug­gest­ing to place air puri­fiers in sub­jects homes, thereby giv­ing them an en­vi­ron­ment that is much lower in PM2.5 for some por­tion of the day, and nor­mal lev­els for the rest of the day. The effects of this on hu­man health might be quite differ­ent than sim­ply ex­pos­ing them to a con­stant ex­po­sure, even if the av­er­ages are the same. The effects might be more pos­i­tive than the calcu­la­tion based on av­er­ages would sug­gest. [5]

Fi­nally, out­side PM2.5 lev­els can vary sig­nifi­cantly de­pend­ing on weather and sea­son. Per­haps any anal­y­sis based on av­er­age PM2.5 lev­els will always paint an in­com­plete pic­ture.

Lin­ear­ity Assumption

Be­cause of the great un­cer­tain­ties men­tioned above, we will as­sume lin­ear scal­ing of health effects based on av­er­age PM2.5 re­duc­tions. We will only use data from one source. That source in­di­cates 0.22 days of life lost from ex­tra ex­po­sure of 1µg/​m³ for one year, it does not take into ac­count dis­abil­ity. We can be fairly con­fi­dent in these num­bers be­cause they are roughly con­sis­tent with the sources from the UK and In­dia.

Air puri­fiers with HEPA filters for PM2.5 mitigation

A stan­dalone air fil­ter­ing de­vice, us­ing a HEPA (high-effi­ciency par­tic­u­late air) filter, can re­duce in­door PM2.5 lev­els sig­nifi­cantly:

Calcu­lat­ing the achiev­able re­duc­tion in per­sonal exposure

To es­ti­mate the benefits of air puri­fier use, we need to know the achiev­able re­duc­tion in per­sonal PM2.5 dose—which in­cludes peo­ple go­ing about their daily lives and not spend­ing all day at home next to the de­vice.

Terms used here:

  • “Per­sonal baseline ex­po­sure”: av­er­age PM2.5 con­cen­tra­tion that a per­son en­coun­ters while go­ing about their daily life, in a va­ri­ety of in­door and out­door set­tings, with­out us­ing air purifiers

  • “Per­sonal miti­gated ex­po­sure”: av­er­age PM2.5 con­cen­tra­tion that a per­son en­coun­ters while go­ing about their daily life, while us­ing air puri­fiers at their home only (not at their work­place or other lo­ca­tions)

In or­der to calcu­late the re­duc­tion in per­sonal ex­po­sure, we will make a sim­ple calcu­la­tion based on some as­sump­tions. Then, we will com­pare it with real-world data from stud­ies in which air puri­fiers were placed in par­ti­ci­pants homes and per­sonal ex­po­sure was mea­sured us­ing portable de­vices.

Step 1: Sim­ple Model

We as­sume an air puri­fier is used in the main bed­room only, and that the bed­room is oc­cu­pied for 10 hours each day. We also as­sume that the win­dows are closed and that the room is fairly small—there­fore we will use the more op­ti­mistic es­ti­mate of a 72% re­duc­tion in PM2.5 ver­sus baseline con­di­tions (refer­enced pre­vi­ously). This yields a to­tal re­duc­tion in per­sonal ex­po­sure of 30%, or a ra­tio be­tween per­sonal miti­gated and per­sonal baseline of 0.7.

Step 2: Com­par­i­son with real-world per­sonal ex­po­sure data

We’ll look at stud­ies with peo­ple car­ry­ing portable mea­sure­ment de­vices, and com­pare them with our calcu­la­tion of a 0.7 ra­tio:

Health Benefits de­liv­ered by this reduction

With the re­duc­tion ra­tio calcu­lated above, we can calcu­late the health benefits. As de­scribed above, we will as­sume a lin­ear re­la­tion­ship be­tween av­er­age PM2.5 ex­po­sure and health effects.

From here on­wards, we will con­sider two lo­ca­tions as ex­am­ples: Sch­wechat, a sub­urb of Vienna, with an av­er­age of 15µg/​m³ PM2.5, and Muzaf­far­naga, In­dia with an av­er­age of 89µg/​m³ PM2.5 (vales from IQair.com). For the health effects, we use the val­ues from the EU and In­dia stud­ies listed un­der “ex­ist­ing work” above.

Cost-Effectiveness

De­vice and Filter Cost

De­vice Cost

Air puri­fiers can cost as lit­tle as $80. The stud­ies cited above demon­strate that even cheap de­vices are suffi­cient for good re­sults. For de­vice life­time, we as­sume 10 years.

Filter Re­place­ment Cost

Perfor­mance of the air puri­fier de­pends on the con­di­tion of the HEPA filter—it needs to be re­placed reg­u­larly, de­pend­ing on the amount of par­ti­cles the filter has already re­moved. I could not find spe­cific in­for­ma­tion on how long a HEPA filter lasts, most sources say “about one year”. A test in Beijing re­vealed a −50% effec­tive­ness drop af­ter 200 days of use, which would mean one year might be too op­ti­mistic in high-pol­lu­tion situ­a­tions. For our two sce­nar­ios of In­dia and Aus­tria, we will as­sume filter life­times of 0,5 years and 1,5 years, re­spec­tively.

A HEPA filter costs around $10. It is pos­si­ble to buy cheaper ones for $4.25 but they perform sig­nifi­cantly worse.

Elec­tric­ity Cost

Air puri­fiers might use 30-50 Watts (source). In our model, we use lo­cal res­i­den­tial elec­tric­ity prices of our two sam­ple lo­ca­tions, In­dia and Aus­tria.

Results

Put­ting these num­bers to­gether, we ar­rive at $5230 per DALY for In­dia and $15200 per DALY for Aus­tria (model here).

Plac­ing these num­bers in con­text:

  • The Against Malaria Foun­da­tion is able to pre­serve a year of healthy life for $78 - two or­ders of mag­ni­tude bet­ter.

  • The WHO con­sid­ers an in­ter­ven­tion that costs, per dis­abil­ity-ad­justed life year (DALY) avoided less than 3x the na­tional an­nual GDP per cap­ita “effec­tive”. An in­ter­ven­tion that costs less than 1x that amount is con­sid­ered “highly effec­tive”. Un­der these crite­ria, the in­ter­ven­tion would be con­sid­ered “highly effec­tive” in Aus­tria and “effec­tive” in In­dia.

For mul­ti­ple peo­ple liv­ing in the same space, cost would go down ac­cord­ingly. For ex­am­ple, for a bed­room shared by two peo­ple, cost per DALY would halve. For a five-per­son fam­ily shar­ing a small flat, with the puri­fier placed in the main liv­ing area, cost effec­tive­ness might be even bet­ter.

Ways to im­prove the cost-effectiveness

  1. Homemade de­vices: An air puri­fier is ba­si­cally just a fan and a pas­sive, re­place­able filter. It is pos­si­ble to build a perfectly ad­e­quate air puri­fier with just that. Some­one has tested this setup in China and got­ten very favourable mea­sure­ments. How­ever, since good air puri­fiers are available for $80, home­made de­vices may not change the calcu­lus much.

  2. Cheaper HEPA filters. How­ever: ac­cord­ing to this blog­post, none ex­ist that are much cheaper yet perform well.

  3. Tar­get­ing the right peo­ple: By in­stal­ling de­vices in the homes of peo­ple who suffer from chronic res­pi­ra­tory dis­ease and who stay home more than the av­er­age, we could some­what in­crease the health benefits de­liv­ered.

  4. Timing: Pol­lu­tion varies sig­nifi­cantly de­pend­ing on weather and sea­son. Cost sav­ings might be achieved by us­ing the air puri­fier only on days of very high pol­lu­tion. This is ques­tion­able though, be­cause of the sub­lin­ear scal­ing of health effects and be­cause HEPA filters de­grade based on the amount of par­ti­cles trapped (so us­ing them only on high-pol­lu­tion days would not re­duce filter cost by much).

  5. Lo­ca­tion: By plac­ing stan­dalone air puri­fiers in offices or schools, sig­nifi­cantly bet­ter cost-effec­tive­ness might be achieved.

  6. In­te­gra­tion with ven­tila­tion sys­tems: Ven­tila­tion sys­tems of pub­lic build­ings could suck all in­com­ing air through HEPA filters. This is already done in some places, but not ev­ery­where. I’m un­sure how much bet­ter the cost-effec­tive­ness would be com­pared to stan­dalone de­vices.

  7. Place­ment: A HEPA filter can re­move more than 99.9% of par­ti­cles. There­fore, out­go­ing air from air puri­fiers is al­most com­pletely free from any par­ti­cles (https://​​www.tand­fon­line.com/​​doi/​​full/​​10.1080/​​02786826.2016.1197375). If some­one were to sit or sleep di­rectly in front of the de­vice, it’s plau­si­ble PM2.5 lev­els could be re­duced down to al­most zero. Pos­si­bly a low-pow­ered air puri­fier, di­rected at a per­sons face while they sleep or work, might de­liver very large re­duc­tions in PM2.5 while re­quiring very lit­tle power. In the sum­mer­time, many peo­ple use fans for per­sonal cool­ing—a very cheap in­ter­ven­tion would be to strap HEPA filters to those ven­tila­tors. This will nat­u­rally re­duce the cool­ing effects, but it might be a very cost-effec­tive way to roll out air purifi­ca­tion. Fur­ther­more, HEPA filters could be made manda­tory for air in­lets in cars.

  8. Power con­sump­tion could be im­proved by build­ing a more en­ergy-effi­cient de­vice.

Virus removal

HEPA filters are good at re­mov­ing very small par­ti­cles. This in­cludes viruses like SARS-CoV-2. It is un­clear whether this would have any sig­nifi­cant pro­tec­tive effect in prac­tice, if a house­hold is shared with an in­fected per­son.

Re­moval of viruses, bac­te­ria and spores might yield ad­di­tional pos­i­tive health effects not already con­sid­ered in our calcu­la­tion. Pos­si­bly a long-term large-scale air puri­fier study would be re­quired to mea­sure these effects.

Nega­tive side effects of this intervention

  • An air puri­fier might in­still a sense of false safety, for ex­am­ple it might lead peo­ple to be­lieve the de­vice offers com­plete pro­tec­tion against sec­ond-hand smoke.

  • Noise con­cerns. In our sam­ple calcu­la­tion above, we as­sumed the air puri­fier is placed in the main bed­room, which could pre­sent prob­lems for sleep qual­ity. I haven’t looked into how loud air puri­fiers are. It is cer­tainly pos­si­ble to build very quiet air puri­fiers, be­cause very quiet fans ex­ist, and HEPA filters don’t need par­tic­u­larly high pres­sure.

Open Questions

… about air pol­lu­tion in general

  • Which other in­ter­ven­tions against air pol­lu­tion could be much more cost-effec­tive? (It seems plau­si­ble that the re­moval of pol­lu­tion sources would be much more im­pact­ful—but also more difficult to achieve—than air fil­ter­ing)

  • How do the cost-effec­tive­ness num­bers calcu­lated here com­pare with ex­ist­ing air qual­ity in­ter­ven­tions and or­ga­ni­za­tions, like the Clean Air Task Force?

… on health effects

  • How can we quan­tify the health benefits of a given re­duc­tion in PM2.5 more pre­cisely? The main challenge is the pat­tern of ex­po­sure: al­ter­nat­ing be­tween low and high. It is plau­si­ble to as­sume that this pat­tern leads to differ­ent health effects than a calcu­la­tion based on av­er­ages would sug­gest.

  • What is the rel­a­tive health im­pact of each class of par­ti­cle sizes? How much of that im­pact comes from ul­tra­fine par­ti­cles? HEPA filters might re­move ul­tra­fine par­ti­cles even bet­ter than they do par­ti­cles with sizes around 1µm. If ul­tra­fine par­ti­cles are what ac­tu­ally causes the worst health effects, it’s plau­si­ble that the ac­tual health benefits are greater than what PM2.5 mea­sure­ments would in­di­cate.

  • Are there stud­ies look­ing at health effects of long-term air puri­fier use? Some stud­ies mea­sure in­di­ca­tors of car­dio­vas­cu­lar health af­ter a short time of air puri­fier use. Can these num­bers be used to es­ti­mate the long-term benefit?

… on the prac­ti­cal­ities of air purifiers

  • In which other ways could the use of air puri­fiers be sup­ported with­out di­rect sub­ven­tion? (for ex­am­ple, policy in­ter­ven­tions)

  • There are many po­ten­tial im­prove­ments to the baseline cost-effec­tive­ness of this in­ter­ven­tion. What kind of effec­tive­ness could be achieved in a best-case sce­nario?

  • Are there or­ga­ni­za­tions lob­by­ing for air puri­fiers already? (apart from the man­u­fac­tur­ers of those de­vices)

  • Would the use of air puri­fiers in schools and office build­ings be more or less cost-effec­tive than the es­ti­mates for homes?

  • What’s the best source for cheap HEPA filters and how cheap would that be?

  • Which ex­ist­ing air puri­fier de­vice is cheap­est, when tak­ing all op­er­at­ing costs into ac­count? And does it plau­si­bly last 10 years or longer?

  • What is the lifes­pan of a HEPA filter, as a func­tion of hours of use and pol­lu­tion lev­els?

  • Would it be pos­si­ble to cre­ate an air puri­fier that’s sig­nifi­cantly cheaper to buy and op­er­ate than ex­ist­ing mod­els? (build­ing an air puri­fier is very easy: sim­ply put a HEPA filter on a fan, as ex­plained in this blog­post). By the way: The blog­post was writ­ten by Thomas Talhelm, who later founded Smart Air, an air puri­fier so­cial en­ter­prise. His blog and com­pany web­site are ex­cel­lent sources for in-depth dis­cus­sion of air pol­lu­tion and filter en­g­ineer­ing. Their de­vice (the “Sqair”) might be a good can­di­date for a low-cost, en­ergy-effi­cient and effec­tive air puri­fier.

  • How would an air puri­fier need to be de­signed in or­der to be easy to use, mostly au­to­matic, and de­liver the in­ter­ven­tion re­li­ably? Does it need to have a PM2.5 sen­sor on­board? How would that af­fect the re­al­is­ti­cally achiev­able cost?

Con­clu­sions and Fur­ther Work

Air pol­lu­tion is one of the biggest pub­lic health prob­lems of our time. Sim­ple air puri­fiers are sur­pris­ingly effec­tive at re­duc­ing the harm. In our sam­ple calcu­la­tion, the in­ter­ven­tion eas­ily meets WHO crite­ria for a “highly effec­tive” in­ter­ven­tion in Aus­tria, and the crite­ria for an “effec­tive” in­ter­ven­tion in In­dia. With just a few small im­prove­ments to cost-effec­tive­ness, it would qual­ify as “highly effec­tive” in In­dia too.

There are many ways in which effec­tive­ness could be im­proved: If the bed­room is shared by two peo­ple, effec­tive­ness dou­bles. Our calcu­la­tions were made for 10 hours per day of use. Many peo­ple stay home for longer than that, and would cor­re­spond­ingly benefit more from an air puri­fier in their home. It is plau­si­ble that we could find more en­ergy-effi­cient de­vices and op­ti­mize lo­ca­tion, place­ment and timing. Fur­ther­more, de­vices could be prefer­en­tially given to in­di­vi­d­u­als which are at spe­cial risk of pol­lu­tion-in­duced ill­ness.

Buy­ing air puri­fiers for peo­ple to place in their homes is prob­a­bly not a promis­ing EA in­ter­ven­tion: Cost-effec­tive­ness is two or­ders of mag­ni­tude worse than GiveWell-recom­mended char­i­ties. That be­ing said, there might be much more cost-effec­tive ways of helping peo­ple get ac­cess to air puri­fiers. We might lobby gov­ern­ments to sub­si­dize those de­vices, or to make HEPA filters manda­tory for pub­lic build­ings and ve­hi­cles.

Beyond air puri­fiers, we could prob­a­bly find other in­ter­ven­tions for miti­gat­ing air pol­lu­tion that are sig­nifi­cantly more cost-effec­tive.

I’ve been quite sur­prised by the re­sults. It seems that us­ing an air puri­fier has solid health benefits, both in very pol­luted and in av­er­agely pol­luted lo­ca­tions. It is sur­pris­ing that in af­fluent coun­tries, where peo­ple can eas­ily af­ford these de­vices, air puri­fier use is not com­mon­place. The health benefits are clear and well-stud­ied. I have in­stalled a home­made de­vice in my bed­room, to­gether with a PM2.5 sen­sor, and plan to place a sec­ond de­vice in the office.

If you’re in­ter­ested in air pol­lu­tion, air puri­fiers, or would like to col­lab­o­rate on fu­ture re­search please get in touch.

Acknowledgements

For their com­ments and feed­back, I’d like to thank An­drés Gómez Emils­son (who pre­vi­ously men­tioned HEPA filters in his post on Cause X), Boyang Xia, Cameron Earl, Ger­not Oh­ner, Han­nah Met­zler, He­lene Kortschak, Lorenz Krüger, Matthew Dahlhausen and Matthias Samwald.


  1. De­tails on HEPA filter effi­ciency on var­i­ous par­ti­cle sizes can be found here: 1, 2 and a sim­ple graph can be found on Wikipe­dia. ↩︎

  2. Dis­abil­ity-ad­justed life years, con­sist­ing of years of life lost plus a weighted sum of years spent in dis­abil­ity. ↩︎

  3. With a life ex­pec­tancy in In­dia of 69 years, this works out to 1.6*365/​69/​89 = 0.095 days of life lost per year and per ex­tra µg/​m³. When com­pared with the EU val­ues above, which are for a PM2.5 range that is much lower than in In­dia, this demon­strates sub­lin­ear scal­ing of health effects at higher lev­els. ↩︎

  4. In or­der to es­ti­mate years of life lost, the ex­tra mor­tal­ity risk is not enough. We would need two ad­di­tional pieces of in­for­ma­tion: The age dis­tri­bu­tion in the pop­u­la­tion, and the ex­tra mor­tal­ity risk per age group ↩︎

  5. We have seen that re­duc­ing av­er­age ex­po­sure by x% re­duces health effects by less than x%, be­cause of the sub­lin­ear scal­ing. How­ever, in our pro­posed in­ter­ven­tion, we’re not only re­duc­ing av­er­age ex­po­sure, we’re also shift­ing the dis­tri­bu­tion of ex­po­sure over time: The hours spent at home will be in an en­vi­ron­ment of greatly re­duced PM2.5, the hours spent el­se­where will be at un­changed ex­po­sure lev­els. If we as­sume that health effects ac­cu­mu­late lin­early hour af­ter hour and that the sub­lin­ear­ity of effects ap­plies to each hour in­di­vi­d­u­ally, this might mean that an av­er­age re­duc­tion of x%, de­liv­ered in this way, would re­duce health effects by more than x%. ↩︎