Burnout: What is it and how to Treat it.

Introduction

Lately there has been con­sid­er­able con­cern among effec­tive al­tru­ists about burnout. Peo­ple are wor­ried about them­selves or oth­ers be­ing less pro­duc­tive* or just plain mis­er­able. The goal of this re­port is to bring peo­ple up to speed on the sci­en­tific re­search about burnout and, when pos­si­ble, make recom­men­da­tions about alle­vi­a­tion and pre­ven­tion. Un­for­tu­nately, the sci­en­tific liter­a­ture has few spe­cific recom­men­da­tions to make, so I would like to use this as an op­por­tu­nity to foster dis­cus­sion about what has worked and not worked for peo­ple per­son­ally. Look for those com­ments be­low. The goal is for this to be use­ful to in­di­vi­d­ual work­ers in treat­ing their own burnout, and to or­ga­ni­za­tional de­ci­sion mak­ers in pre­vent­ing burnout or­ga­ni­za­tion-wide.

[*Stud­ies are ac­tu­ally mixed on if burnout re­duces pro­duc­tivity, with some even show­ing burnout as­so­ci­ated with higher pro­duc­tivity. My in­ter­pre­ta­tion here is that high stan­dards for your­self lead to both high perfor­mance and burnout.]

Tl;dr

  1. So­cial sup­port == Good.

  2. Sleep == Good.

  3. Am­bi­guity == Bad.

  4. Va­ca­tions == Meh.

What is Burnout?

The offi­cial defi­ni­tion of burnout is “phys­i­cal or men­tal col­lapse caused by over­work or stress”. That kind of im­plies that a per­son can’t work when burnt out, but that’s not my ex­pe­rience- ceas­ing work when you’re burnt out is a priv­ilege. But work­ing when you’re burnt out is mis­er­able, and makes burnout worse, so even if cir­cum­stances im­prove you’re in a hole.

Burnout was origi­nally con­ceived of in the car­ing pro­fes­sions (e.g. nurs­ing and so­cial work), which are emo­tion­ally de­mand­ing in sev­eral differ­ent ways. This has by and large not been born out sci­en­tifi­cally; other pro­fes­sions burn out just as hard, with per­haps slightly differ­ent pat­terns on the Maslach Burnout In­ven­tory, the most pop­u­lar mea­sure of burnout. Based mostly on per­sonal ob­ser­va­tion I strongly sus­pect there are mul­ti­ple types of burnout, which can co-oc­cur, and which cur­rent in­stru­ments are not sen­si­tive enough to differ­en­ti­ate. Of par­tic­u­lar in­ter­est to this crowd is the differ­ence be­tween burnout caused by hat­ing your job or not hav­ing the re­sources it de­mands, vs. lov­ing your job too much and be­ing sucked into giv­ing more than you should. I sus­pect that the lat­ter is more heav­ily rep­re­sented among effec­tive al­tru­ists than in the liter­a­ture.

The Maslach Burnout In­ven­tory (used in >90% of stud­ies) di­vides burnout into three parts: ex­haus­tion, cyn­i­cism, and (per­ceived) per­sonal effi­cacy. The MBI has been shown to be in­ter­nally con­sis­tent and cross-cul­turally valid. On the other hand, it has mixed re­sults in dis­t­in­guish­ing burnout from tra­di­tional de­pres­sion or anx­iety, and I could find no stud­ies demon­strat­ing any pre­dic­tive value of the in­ven­tory — the clos­est was two stud­ies show­ing MBI pre­dicted an in­crease in thoughts of suicide and drop­ping out of school among med stu­dents.

In con­trast, the Copen­hagen Burnout In­ven­tory has one whole study show­ing a that a high score pre­dicts fu­ture sick­ness ab­sence, sleep prob­lems, and use of painkil­lers. The CBI mea­sures only ex­haus­tion, and sep­a­rately tracks per­sonal burnout, work burnout, and client burnout. I would have liked to give prefer­ence to stud­ies us­ing the CBI be­cause it has more em­piri­cal val­i­da­tion, but there sim­ply weren’t enough to rely on, so most of the stud­ies referred to in this post use the MBI.

By far the most pop­u­lar model of burnout in the liter­a­ture is “Job De­mands—Re­sources”, or “JD-R”, which posits that high job de­mands lead to ex­haus­tion, and low re­sources lead to cyn­i­cism and feel­ings of low per­sonal effi­cacy. “De­mands” and “Re­sources” are defined fairly broadly here. De­mands in­cludes things like “cop­ing with con­flict­ing goals” and re­sources in­cludes things like chances for ad­vance­ment.

There is a similar the­ory called Con­ser­va­tion of Re­sources. It has less em­piri­cal sup­port and does not make no­tice­ably differ­ent pre­dic­tions, so I won’t pay it fur­ther at­ten­tion.

What Can You do as an In­di­vi­d­ual?

An un­for­tu­nate fact is that the more priv­ileged you are, the eas­ier it is to cre­ate work­place that ac­com­mo­dates you. Th­ese sug­ges­tions will be eas­ier to im­ple­ment for peo­ple in higher de­mand pro­fes­sions, who earn more money, or have bet­ter bosses.

A sec­ond un­for­tu­nate fact is that the liter­a­ture was not very helpful in recom­mend­ing ways to pre­vent or re­lieve burnout. The fol­low­ing are ei­ther based on in­ter­pre­ta­tions of liter­a­ture (e.g. be­cause so­cial sup­port is cor­re­lated with low burnout, I recom­mend so­cial sup­port) or my own be­liefs based on what I’ve ob­served per­son­ally.

I’ll cre­ate threads for all of these seek­ing anec­dotes about how you’ve achieved them for your­self in the com­ments.

Seek So­cial Support

The most con­sis­tent find­ing in all burnout liter­a­ture is that so­cial sup­port (lead­er­ship and peer) re­duces burnout. That is not even all the stud­ies I found on this point.

Sleep

Un­sur­pris­ingly, sleep du­ra­tion is nega­tively cor­re­lated with burnout. I think the causal­ity here is up for de­bate — stress causes poor sleep in par­allel to poor sleep caus­ing de­pres­sion — but to the ex­tent you can im­prove your sleep, it’s likely to be helpful.

Seek Clar­ity in Your Role and Goals

Am­bi­guity (or worse, con­flict­ing goals) leads to in­ter­nal con­flict, which leads to burnout.

Shorten Your Commute

Com­mut­ing is an ab­solutely mis­er­able ex­pe­rience. It is one of few mis­eries hu­mans can’t ac­cli­mate to. It cuts into sleep and per­sonal time. Do what­ever you can to shorten your com­mute.

Buy Your­self Out of Work-Life Conflict

This is a more gen­eral ap­pli­ca­tion of “don’t com­mute”.

Of course, this ad­vice is only ac­tion­able if you have money. How­ever there is a no­tice­able con­tin­gent within EA of peo­ple push­ing them­selves to live on much, much less than they earn, and it’s worth ex­am­in­ing whether this is penny-wise-pound-fool­ish for you per­son­ally. Buy­ing rest might re­duce your dona­tions in the short term, but if it in­creases your long term ca­pac­ity, it is the right thing to do.

Keep a Healthy Per­sonal Runway

80,000 hours goes into the mul­ti­tude of rea­sons you should do this here, but I want to ex­plic­itly call out sav­ings’ role in fight­ing burnout. Know­ing you can leave can en­able you to de­mand bet­ter treat­ment. Know­ing you don’t need a par­tic­u­lar job lets you hold out for a bet­ter fit.

No­tice­able Ab­sence: Vacations

Va­ca­tion is com­monly con­sid­ered the treat­ment for burnout, how­ever one of the most con­sis­tent find­ings in my re­search is that the re­duc­tion in burnout fol­low­ing a va­ca­tion dis­si­pates very quickly — in less than three weeks.

This is one of those times I feel a strug­gle be­tween what I “know” anec­do­tally, what re­search demon­strates, and my per­sonal ex­pe­rience. I’ve seen peo­ple in very de­mand­ing jobs go on va­ca­tion, clear their head, and come back think­ing more strate­gi­cally. But this has never worked for me per­son­ally: when a job is un­pleas­ant, it’s un­pleas­ant, and leav­ing for a week doesn’t make it more pleas­ant. I con­jec­ture that there are re­ally two forms of burnout — “this job is ter­rible” and “this great job is very de­mand­ing”, and there sim­ply aren’t enough of the lat­ter to show up in stud­ies. This is un­for­tu­nate, given that I ex­pect it to be over­rep­re­sented among peo­ple read­ing this. But at a min­i­mum, you should not count on va­ca­tion to fix a bad work en­vi­ron­ment.

What Can Or­ga­ni­za­tions do?

Ul­ti­mately burnout is pre­vented by in­di­vi­d­u­als, not com­pa­nies. The warmest, fuzziest com­pany in the world can’t pre­vent peo­ple from work­ing them­selves into burnout if they’re truly de­ter­mined. At the same time, or­ga­ni­za­tions are set­ting the in­cen­tive struc­tures that de­ter­mine whether pre­vent­ing burnout is an up­hill fight or not, and I think this makes them the cor­rect place for most in­ter­ven­tions.

Clear Roles and Norms

Am­bi­guity as to goals or so­cial norms leads to stress and thus burnout.

Clear Feedback

Un­for­tu­nately noth­ing I read went into what con­sti­tuted good feed­back or how to give it. How­ever in gen­eral re­duc­ing am­bi­guity is good and re­duces burnout.

Achiev­able Goals

It has be­come pop­u­lar to set “stretch” goals. Google’s quar­terly eval­u­a­tion pro­cess says that if you reg­u­larly achieve more than 60-70% of your goals, you’re aiming too low. This has the ad­van­tage of never leav­ing some­one un­der ca­pac­ity, but also means they never get to feel like they’ve “won”, which leads to burnout.

As a bonus, keep­ing goals mod­est leaves peo­ple with enough re­serves and ca­pac­ity to sprint when you re­ally need it.

Fa­cil­i­tate So­cial Support

Feel­ing so­cially sup­ported is one of the few demon­strated pro­tec­tors against burnout. The liter­a­ture is mixed at de­scribing spe­cific in­ter­ven­tions — some stud­ied de­liber­ate in­ter­ven­tions like peer sup­port groups, oth­ers merely counted the num­ber of pos­i­tive and nega­tive in­ter­ac­tions. I re­port the peer re­view sup­ported in­ter­ven­tions be­low, and leave the anec­dotes to the com­ments.

  • Peer sup­port groups (stud­ied only in client-fac­ing, emo­tion­ally de­mand­ing pro­fes­sions like nurs­ing and so­cial work, my in­stinct is that this will not trans­fer di­rectly to office po­si­tions)

  • Strong on­board­ing pro­grams that make peo­ple feel so­cially situ­ated (as well as giv­ing them clear goals, and the re­sources to do their job).

  • Ex­plicit men­tor­ship pro­grams (which can also re­duce am­bi­guity).

A warn­ing: work­ers of­ten rec­og­nize at­tempts to fa­cil­i­tate group co­he­sion and feel obliged to fake feel­ings of con­nec­tion when they can’t feel it au­then­ti­cally. My be­lief is that this de­mand for emo­tional la­bor can make burnout worse. You can miti­gate this by pro­vid­ing a bond­ing bud­get for the em­ploy­ees them­selves choose how to spend — al­though even then, not ev­ery­one will love the same things, and peo­ple may feel pres­sured to par­ti­ci­pate or left out if they don’t. You can also make ex­plicit in the in­ter­view what kind of so­cial ex­pec­ta­tions you have, and let peo­ple who aren’t com­pat­i­ble opt out.

Fa­cil­i­tate Telecom­mut­ing and Flex­ible Work Hours

In a re­sult that is very grat­ify­ing to me per­son­ally, re­mote work and flex time gen­er­ally re­duce burnout. How­ever it can’t be done blindly. My per­sonal ex­pe­rience has been that if a com­pany’s work­flow is set up for on-site work (e.g. no one re­sponds to IM and you have to phys­i­cally track peo­ple down to get an­swers, in­for­ma­tion is gen­er­ally not writ­ten down), re­mote work re­duces per­sonal effi­cacy to the point that it’s not worth it. But when it is sup­ported (mean­ing, co-work­ers are re­spon­sive, a strong cul­ture of writ­ing things down), telecom­mut­ing re­moves com­mut­ing, re­duces work-life con­flict, and gives peo­ple more con­trol over their en­vi­ron­ment.

As a bonus, re­mote work makes it im­pos­si­ble to know how many hours peo­ple are work­ing, re­mov­ing the in­cen­tive for pro­duc­tivity the­ater.

Two caveats to this:

  • Re­mote work is bet­ter for in­tro­verts than ex­tro­verts. Some ex­tro­verts may hate it.

  • After a cer­tain point (this meta-anal­y­sis says half the week), re­mote work weak­ens so­cial bonds, which is shown el­se­where to in­crease burnout. I be­lieve this can be ame­lio­rated with things like a strong Slack cul­ture and reg­u­lar re­treats. I felt more sup­ported and con­nected in my time at Wave (an en­tirely re­mote com­pany) than I did at any pre­vi­ous job.

Fa­cil­i­tate Sleep

I found no peer re­viewed study on this, but firmly be­lieve nap rooms to be helpful when prop­erly im­ple­mented. A note to Google: in a busy hal­lway next to the noisy mas­sage chair does not count as proper im­ple­men­ta­tion.

Espe­cially in in­ter­na­tional com­pa­nies, peo­ple can be pres­sured to at­tend meet­ings both very early and very late. I don’t have a good solu­tion to this as long as time zones are real.

Pri­vate Offices, or at Least Cubicles

At this point the ev­i­dence is over­whelming that most peo­ple hate, are less pro­duc­tive and more sick in open offices. Open offices don’t even in­crease in per­son in­ter­ac­tions be­tween co-work­ers. Plus you can nap in pri­vate offices.

Autonomy

Au­ton­omy makes peo­ple happy, and lack of it leads to cyn­i­cism and feel­ings of low per­sonal effi­cacy (it doesn’t ap­pear to have much effect on ex­haus­tion).

Au­ton­omy can con­flict some­what with hav­ing a clearly defined role and goal, and the cor­rect bal­ance will vary across peo­ple.

Find Ways to Mea­sure Pro­duc­tivity that Aren’t Hours Worked

If you mea­sure hours visi­bly at work, peo­ple will Good­hart and put in more hours at the office. This may not make them more pro­duc­tive, and de­pend­ing on who you ask may not be able to make them more pro­duc­tive- hu­mans only have so many hours of thought-work in them per day. But it will eat up their time and pre­vent them from re­cov­er­ing.

Pro­fes­sion­al­ize Management

Like a lot of young or­ga­ni­za­tions, EA orgs tend to have man­agers who were hired for things other than man­age­ment, have not re­ceived ex­plicit man­age­ment train­ing, and have pri­ori­ties other than man­age­ment. This leads to a lot of pre­dictable prob­lems- they don’t del­e­gate well, don’t give the right feed­back at the right times, etc. “Learn­ing to man­age” would lead to more work ac­com­plished and a bet­ter work en­vi­ron­ment for ev­ery­one. Un­for­tu­nately ev­ery­one else is mak­ing this mis­take too, and good re­sources for learn­ing to man­age are thin on the ground. But a good start would be to sim­ply rec­og­nize man­age­ment as what you are do­ing and a skill that needs to be learned.

What’s Miss­ing?

There were a num­ber of top­ics I hoped to cover but couldn’t, due to a lack of data.

  • Warn­ing signs of burnout

  • Re­cov­ery techniques

  • How to dis­t­in­guish be­tween burnout and similar look­ing prob­lems, like de­pres­sion and anx­iety.

Thanks to CEA for com­mis­sion­ing and fund­ing this re­search.

If you are ex­cep­tion­ally cu­ri­ous about my re­search, you can view my notes here.