Affective disorders

Among the studies on affective disorders (Fig. 2(a)), 31 found increases during the COVID-19 pandemic^{Reference Meda, Pardini, Slongo, Bodini, Zordan and Rigobello14–Reference Elmer, Mepham and Stadtfeld44} and two found increases in subgroups of participants.^{Reference Janssen, Kullberg, Verkuil, van Zwieten, Wever and van Houtum45,Reference Gallagher, Bennett and Roper46} These were conducted on population-based samples (151–336 52 participants);^{Reference Ettman, Abdalla, Cohen, Sampson, Vivier and Galea17,Reference Daly, Sutin and Robinson18,Reference Wanberg, Csillag, Douglass, Zhou and Pollard22,Reference Kwong, Pearson, Adams, Northstone, Tilling and Smith29,Reference Zhao, Wong, Luk, Wai, Lam and Wang33,Reference Creese, Khan, Henley, O'Dwyer, Corbett and Vasconcelos Da Silva39,Reference Twenge and Joiner40,Reference Peters, Rospleszcz, Greiser, Dallavalle and Berger43} more specific healthy populations of various ages, life stages or occupations (93–7527 participants);^{Reference Meda, Pardini, Slongo, Bodini, Zordan and Rigobello14,Reference Krendl and Perry15,Reference Gallagher and Wetherell21,Reference Hamadani, Hasan, Baldi, Hossain, Shiraji and Bhuiyan23–Reference Saraswathi, Saikarthik, Senthil Kumar, Madhan Srinivasan, Ardhanaari and Gunapriya27,Reference Huckins, daSilva, Wang, Hedlund, Rogers and Nepal34,Reference Li, Cao, Leung and Mak36–Reference Creese, Khan, Henley, O'Dwyer, Corbett and Vasconcelos Da Silva39,Reference Zhang, Xiang and Alejok41,Reference Zhang, Zaman, Silenzio, Kautz and Hoque42} and patients/populations with various somatic or psychiatric diagnoses (46–1 854 742 participants).^{Reference Pan, Kok, Eikelenboom, Horsfall, Jörg and Luteijn16,Reference Puhl, Lessard, Larson, Eisenberg and Neumark-Stzainer19,Reference Villani, Vetrano, Damiano, Paola, Ulgiati and Martin20,Reference Stojanov, Malobabic, Milosevic, Stojanov, Vojinovic and Stanojevic28,Reference Thombs, Kwakkenbos, Henry, Carrier, Patten and Harb30–Reference Titov, Staples, Kayrouz, Cross, Karin and Ryan32,Reference Lim, Woo, Lim, Ng, Chan and Gandhi35} The studies were conducted in Hong Kong,^{Reference Wong, Zhang, Sit, Yip, Chung and Wong26,Reference Zhao, Wong, Luk, Wai, Lam and Wang33} the USA,^{Reference Krendl and Perry15,Reference Ettman, Abdalla, Cohen, Sampson, Vivier and Galea17–Reference Puhl, Lessard, Larson, Eisenberg and Neumark-Stzainer19,Reference Wanberg, Csillag, Douglass, Zhou and Pollard22,Reference Lee, Cadigan and Rhew24,Reference Thombs, Kwakkenbos, Henry, Carrier, Patten and Harb30,Reference Huckins, daSilva, Wang, Hedlund, Rogers and Nepal34,Reference Twenge and Joiner40,Reference Zhang, Zaman, Silenzio, Kautz and Hoque42} the UK,^{Reference Gallagher and Wetherell21,Reference Kwong, Pearson, Adams, Northstone, Tilling and Smith29,Reference Thombs, Kwakkenbos, Henry, Carrier, Patten and Harb30,Reference Huckins, daSilva, Wang, Hedlund, Rogers and Nepal34,Reference Creese, Khan, Henley, O'Dwyer, Corbett and Vasconcelos Da Silva39} Germany,^{Reference Jacob, Smith, Koyanagi, Oh, Tanislav and Shin31,Reference Peters, Rospleszcz, Greiser, Dallavalle and Berger43} China,^{Reference Li, Cao, Leung and Mak36,Reference Chen, Chen, Pakpour, Griffiths and Lin37} Italy,^{Reference Meda, Pardini, Slongo, Bodini, Zordan and Rigobello14,Reference Villani, Vetrano, Damiano, Paola, Ulgiati and Martin20,Reference Zanardo, Manghina, Giliberti, Vettore, Severino and Straface25} Australia,^{Reference Titov, Staples, Kayrouz, Cross, Karin and Ryan32,Reference Magson, Freeman, Rapee, Richardson, Oar and Fardouly38} Bangladesh,^{Reference Hamadani, Hasan, Baldi, Hossain, Shiraji and Bhuiyan23} India,^{Reference Saraswathi, Saikarthik, Senthil Kumar, Madhan Srinivasan, Ardhanaari and Gunapriya27} Switzerland,^{Reference Elmer, Mepham and Stadtfeld44} South Sudan,^{Reference Zhang, Xiang and Alejok41} Canada,^{Reference Thombs, Kwakkenbos, Henry, Carrier, Patten and Harb30} France,^{Reference Thombs, Kwakkenbos, Henry, Carrier, Patten and Harb30} Singapore,^{Reference Lim, Woo, Lim, Ng, Chan and Gandhi35} Serbia^{Reference Stojanov, Malobabic, Milosevic, Stojanov, Vojinovic and Stanojevic28} and The Netherlands.^{Reference Pan, Kok, Eikelenboom, Horsfall, Jörg and Luteijn16} Among these studies, four were of high quality^{Reference Gallagher and Wetherell21,Reference Saraswathi, Saikarthik, Senthil Kumar, Madhan Srinivasan, Ardhanaari and Gunapriya27,Reference Kwong, Pearson, Adams, Northstone, Tilling and Smith29,Reference Jacob, Smith, Koyanagi, Oh, Tanislav and Shin31} and 27 were of fair quality.^{Reference Meda, Pardini, Slongo, Bodini, Zordan and Rigobello14–Reference Villani, Vetrano, Damiano, Paola, Ulgiati and Martin20,Reference Wanberg, Csillag, Douglass, Zhou and Pollard22–Reference Wong, Zhang, Sit, Yip, Chung and Wong26,Reference Stojanov, Malobabic, Milosevic, Stojanov, Vojinovic and Stanojevic28,Reference Thombs, Kwakkenbos, Henry, Carrier, Patten and Harb30,Reference Titov, Staples, Kayrouz, Cross, Karin and Ryan32–Reference Elmer, Mepham and Stadtfeld44} A fair-quality study of adolescents and parents from The Netherlands found increased negative affect only in parents,^{Reference Janssen, Kullberg, Verkuil, van Zwieten, Wever and van Houtum45} and a fair-quality study of people with cancer from the UK found increased rates of depression only among those with certain cancer types.^{Reference Gallagher, Bennett and Roper46}

Fig. 2 Harvest plot for the associations reported between exposure to the COVID-19 pandemic and (a) affective disorders, (b) suicides, (c) other mental health outcomes and (d) healthcare utilisation. Labels on the x-axis refer to the reference list entries for the studies.

Altogether, five studies with more defined samples of various ages, occupations and health conditions, with 25–3983 participants, found no change in affective disorders. They were conducted in Canada^{Reference McArthur, Saari, Heckman, Wellens, Weir and Hebert47} the USA^{Reference Pinkham, Ackerman, Depp, Harvey and Moore48,Reference Sturman49} The Netherlands^{Reference van der Velden, Contino, Das, van Loon and Bosmans50} and Italy.^{Reference Baiano, Zappullo and Conson51} Four of these studies were of fair quality,^{Reference McArthur, Saari, Heckman, Wellens, Weir and Hebert47,Reference Pinkham, Ackerman, Depp, Harvey and Moore48,Reference van der Velden, Contino, Das, van Loon and Bosmans50,Reference Baiano, Zappullo and Conson51} and one was of low quality.^{Reference Sturman49}

Some studies found unchanged or lower rates of affective disorders,^{Reference Pariente, Wissotzky Broder, Sheiner, Lanxner Battat, Mazor and Yaniv Salem52–Reference Li, Yu, Miller, Yang and Rouen55} and lower incidence of medication prescriptions.^{Reference Williams, Jenkins, Ashcroft, Brown, Campbell and Carr56} These were conducted on populations of 164–241 458 participants, including postpartum women in Israel,^{Reference Pariente, Wissotzky Broder, Sheiner, Lanxner Battat, Mazor and Yaniv Salem52} patients in general practice in the UK,^{Reference Williams, Jenkins, Ashcroft, Brown, Campbell and Carr56} medical students from the Republic of Kazakhstan,^{Reference Bolatov, Seisembekov, Askarova, Baikanova, Smailova and Fabbro54} patients from a sleep clinic from Japan^{Reference Ubara, Sumi, Ito, Matsuda, Matsuo and Miyamoto53} and university students in China.^{Reference Li, Yu, Miller, Yang and Rouen55} Three of these studies were of fair quality^{Reference Ubara, Sumi, Ito, Matsuda, Matsuo and Miyamoto53–Reference Li, Yu, Miller, Yang and Rouen55} and two were of high quality.^{Reference Pariente, Wissotzky Broder, Sheiner, Lanxner Battat, Mazor and Yaniv Salem52,Reference Williams, Jenkins, Ashcroft, Brown, Campbell and Carr56}

Suicides

Four studies assessed pandemic-period suicide rates in whole populations from Connecticut (USA),^{Reference Mitchell and Li57} Queensland (Australia),^{Reference Leske, Kõlves, Crompton, Arensman and de Leo58} Japan^{Reference Isumi, Doi, Yamaoka, Takahashi and Fujiwara59} and Peru,^{Reference Calderon-Anyosa and Kaufman60} and found these had either decreased^{Reference Mitchell and Li57,Reference Calderon-Anyosa and Kaufman60} or remained unaltered (Fig. 2(b)).^{Reference Leske, Kõlves, Crompton, Arensman and de Leo58,Reference Isumi, Doi, Yamaoka, Takahashi and Fujiwara59} All four studies were of high quality.

Other mental health outcomes

There were 30 studies that assessed other mental health outcomes (Fig. 2(c)). Altogether, 12 studies were conducted on population-based samples and found decreases in mental health.^{Reference Sutin, Stephan, Luchetti, Aschwanden, Strickhouser and Lee61–Reference Ran, Gao, Lin, Zhang, Chan and Deng72} These studies were conducted on populations ranging from 1003 to 17 452 individuals in the USA,^{Reference Sutin, Stephan, Luchetti, Aschwanden, Strickhouser and Lee61,Reference Twenge and Joiner67,Reference McGinty, Presskreischer, Han and Barry68} the UK,^{Reference Pierce, Hope, Ford, Hatch, Hotopf and John62–Reference Daly, Sutin and Robinson65,Reference Niedzwiedz, Green, Benzeval, Campbell, Craig and Demou69} New Zealand,^{Reference Sibley, Greaves, Satherley, Wilson, Overall and Lee66} Denmark,^{Reference Sonderskov, Dinesen, Santini and Ostergaard70} Canada^{Reference Bierman and Schieman71} and China.^{Reference Ran, Gao, Lin, Zhang, Chan and Deng72} Four of these studies were of high quality,^{Reference Pierce, Hope, Ford, Hatch, Hotopf and John62,Reference Daly, Sutin and Robinson65,Reference Twenge and Joiner67,Reference Niedzwiedz, Green, Benzeval, Campbell, Craig and Demou69} seven were of fair quality^{Reference Sutin, Stephan, Luchetti, Aschwanden, Strickhouser and Lee61,Reference Banks and Xu63,Reference Gray, O'Connor, Knowles, Pink, Simkiss and Williams64,Reference Sibley, Greaves, Satherley, Wilson, Overall and Lee66,Reference McGinty, Presskreischer, Han and Barry68,Reference Bierman and Schieman71,Reference Ran, Gao, Lin, Zhang, Chan and Deng72} and one was of low quality.^{Reference Sonderskov, Dinesen, Santini and Ostergaard70}

Ten other studies in more defined samples ranging from 21 to 3505 individuals also found deteriorations in mental health.^{Reference Shen and Bartram73–Reference Gomez, Anderson, Yu, Gutsche, Jablonski and Martin82} These included populations of different ages and occupations,^{Reference Shen and Bartram73–Reference Savage, James, Magistro, Donaldson, Healy and Nevill76,Reference Copeland, McGinnis, Bai, Adams, Nardone and Devadanam78,Reference Dragun, Veček, Marendić, Pribisalić, Đivić and Cena79,Reference Gomez, Anderson, Yu, Gutsche, Jablonski and Martin82} and patients with various somatic or psychiatric diagnoses,^{Reference Ohliger, Umpierrez, Buehler, Ohliger, Magister and Vallier77,Reference Castellini, Cassioli, Rossi, Innocenti, Gironi and Sanfilippo80–Reference Gomez, Anderson, Yu, Gutsche, Jablonski and Martin82} and were conducted in the USA,^{Reference Ohliger, Umpierrez, Buehler, Ohliger, Magister and Vallier77,Reference Copeland, McGinnis, Bai, Adams, Nardone and Devadanam78,Reference Gomez, Anderson, Yu, Gutsche, Jablonski and Martin82} Spain,^{Reference Reverté-Villarroya, Ortega, Lavedán, Masot, Burjalés-Martí and Ballester-Ferrando74} Switzerland,^{Reference Macdonald and Hülür75} Croatia,^{Reference Dragun, Veček, Marendić, Pribisalić, Đivić and Cena79} the UK^{Reference Shen and Bartram73,Reference Savage, James, Magistro, Donaldson, Healy and Nevill76} and Italy.^{Reference Castellini, Cassioli, Rossi, Innocenti, Gironi and Sanfilippo80,Reference Giordano, Siciliano, De Micco, Sant'Elia, Russo and Tedeschi81} One was of high quality^{Reference Ohliger, Umpierrez, Buehler, Ohliger, Magister and Vallier77} and nine were of fair quality.^{Reference Shen and Bartram73–Reference Savage, James, Magistro, Donaldson, Healy and Nevill76,Reference Copeland, McGinnis, Bai, Adams, Nardone and Devadanam78–Reference Gomez, Anderson, Yu, Gutsche, Jablonski and Martin82}

Eight of the studies did not find changes in mental health among study populations of 46–1870 participants. These populations were of various ages and occupations, both healthy and with somatic or mental health diagnoses, conducted in the USA,^{Reference Breslau, Finucane, Locker, Baird, Roth and Collins83–Reference Benham86} Sweden,^{Reference Kivi, Hansson and Bjälkebring87} Germany^{Reference Schäfer, Sopp, Schanz, Staginnus, Göritz and Michael88} and The Netherlands.^{Reference van Tilburg, Steinmetz, Stolte, van der Roest and de Vries89,Reference van Gorp, Maurice-Stam, Teunissen, van de Peppel-van der Meer, Huussen and Schouten-van Meeteren90} Of these studies, two were of high quality^{Reference Breslau, Finucane, Locker, Baird, Roth and Collins83,Reference van Tilburg, Steinmetz, Stolte, van der Roest and de Vries89} and six were of fair quality.^{Reference Penner, Ortiz and Sharp84–Reference Schäfer, Sopp, Schanz, Staginnus, Göritz and Michael88,Reference van Gorp, Maurice-Stam, Teunissen, van de Peppel-van der Meer, Huussen and Schouten-van Meeteren90}

Healthcare utilisation

Figure 2(d) presents a harvest plot for associations between COVID-19 and healthcare utilisation. Altogether, five studies that assessed admissions for mental health problems found decreases: emergency department presentations decreased at three health services in Australia^{Reference Dragovic, Pascu, Hall, Ingram and Waters91} and two hospitals in Italy;^{Reference Stein, Giordano, Del Giudice, Basi, Gambini and D'Agostino92,Reference Capuzzi, Di Brita, Caldiroli, Colmegna, Nava and Buoli93} psychiatric emergency services presentations decreased in Paris, France;^{Reference Pignon, Gourevitch, Tebeka, Dubertret, Cardot and Dauriac-Le Masson94} and presentations to a paediatric emergency department decreased in the USA.^{Reference Leff, Setzer, Cicero and Auerbach95} All of these five studies were of high quality.

On the other hand, although acute care presentations for mental health diagnoses in the UK decreased, the patients admitted had more severe conditions.^{Reference Abbas, Kronenberg, McBride, Chari, Alam and Mukaetova-Ladinska96} Admissions for mental health problems increased at an acute medical unit,^{Reference Grimshaw and Chaudhuri97} and there was acceleration in urgent referrals to secondary mental health services in the UK.^{Reference Chen, She, Qin, Kershenbaum, Fernandez-Egea and Nelder98} In Italy, psychological morbidity worsened among 145 palliative care professionals.^{Reference Varani, Ostan, Franchini, Ercolani, Pannuti and Biasco99} An emergency department in New Zealand experienced overall decreases in mental health presentations, but relative increases in overdoses and self-harm.^{Reference Joyce, Richardson, McCombie, Hamilton and Ardagh100} Four^{Reference Abbas, Kronenberg, McBride, Chari, Alam and Mukaetova-Ladinska96–Reference Chen, She, Qin, Kershenbaum, Fernandez-Egea and Nelder98,Reference Joyce, Richardson, McCombie, Hamilton and Ardagh100} of these studies were of high quality and one was of fair quality.^{Reference Varani, Ostan, Franchini, Ercolani, Pannuti and Biasco99}

Affective disorders

Figure 3(a) presents a harvest plot for associations between economic crises and affective disorders. All 15 studies reporting affective disorders as an outcome were population-based surveys. The findings from 12 of these studies, with populations ranging from 2011 to 81 313 participants, were that there was a significant increase in affective disorders.^{Reference Wang, Smailes, Sareen, Fick, Schmitz and Patten101–Reference Shi, Taylor, Goldney, Winefield, Gill and Tuckerman112} These studies were conducted in Canada,^{Reference Wang, Smailes, Sareen, Fick, Schmitz and Patten101} Hong Kong,^{Reference Lee, Guo, Tsang, Mak, Wu and Ng102} the USA,^{Reference Riumallo-Herl, Basu, Stuckler, Courtin and Avendano103–Reference Cagney, Browning, Iveniuk and English105,Reference Mehta, Kramer, Durazo-Arvizu, Cao, Tong and Rao108–Reference Dagher, Chen and Thomas111} Europe,^{Reference Riumallo-Herl, Basu, Stuckler, Courtin and Avendano103} Spain^{Reference Chaves, Castellanos, Abrams and Vazquez106} and Australia.^{Reference Sargent-Cox, Butterworth and Anstey107,Reference Shi, Taylor, Goldney, Winefield, Gill and Tuckerman112} Two of these studies were of high quality^{Reference Riumallo-Herl, Basu, Stuckler, Courtin and Avendano103,Reference Tapia Granados, Christine, Ionides, Carnethon, Diez Roux and Kiefe104} and ten were of fair quality.^{Reference Wang, Smailes, Sareen, Fick, Schmitz and Patten101, Reference Lee, Guo, Tsang, Mak, Wu and Ng102, Reference Cagney, Browning, Iveniuk and English105–Reference Shi, Taylor, Goldney, Winefield, Gill and Tuckerman112}

Fig. 3 Harvest plot for the associations reported between exposure to the economic crisis and (a) affective disorders, (b) suicides, (c) other mental health outcomes and (d) healthcare utilisation. Labels on the x-axis refer to the reference list entries for the studies.

A study on 815 adults aged over 50 years found no increase in depression among those most affected by the stock market crash, despite an increase in antidepressant medication use.^{Reference McInerney, Mellor and Nicholas113} Also, a study on 25- to 75-year-olds in the USA found that mental health improved.^{Reference Forbes and Krueger114} Among 106 158 participants aged over 15 years from 21 European countries, no effect of the crisis was found on depressive feelings.^{Reference Reibling, Beckfield, Huijts, Schmidt-Catran, Thomson and Wendt115} One study was of high quality,^{Reference Forbes and Krueger114} and two were of fair quality.^{Reference McInerney, Mellor and Nicholas113,Reference Reibling, Beckfield, Huijts, Schmidt-Catran, Thomson and Wendt115}

Suicides

Altogether, 37 studies assessed suicide in relation to the 2008 economic crisis, and all of these studies were of high quality (Fig. 3(b)). Altogether, 19 studies found increased suicide rates at the level of the total population after the start of the crisis. These were conducted on the populations of Italy (Milan)^{Reference Merzagora, Mugellini, Amadasi and Travaini116} (suicides as a result of mental and behavioural disorders, Italy^{Reference De Vogli, Vieno and Lenzi117}), Greece,^{Reference Zilidis, Papagiannis and Rachiotis118–Reference Kontaxakis, Papaslanis, Havaki-Kontaxaki, Tsouvelas, Giotakos and Papadimitriou123} Spain^{Reference Lopez Bernal, Gasparrini, Artundo and McKee124} (suicide attempts in Spain^{Reference Córdoba-Doña, San Sebastián, Escolar-Pujolar, Martínez-Faure and Gustafsson125}), the European Union,^{Reference Reeves, McKee and Stuckler126–Reference Laanani, Ghosn, Jougla and Rey128} Canada,^{Reference Reeves, McKee and Stuckler126} England,^{Reference Laanani, Ghosn, Jougla and Rey128–Reference Barr, Taylor-Robinson, Scott-Samuel, McKee and Stuckler130} the USA^{Reference Agrrawal, Waggle and Sandweiss131–Reference Carriere, Marshall and Binkley133} and South Korea.^{Reference Chan, Caine, You, Fu, Chang and Yip134}

Some studies reported increases in suicide rates in specific population subgroups,^{Reference Rachiotis, Stuckler, McKee and Hadjichristodoulou135,Reference Alexopoulos, Kavalidou and Messolora136} among men^{Reference Coope, Gunnell, Hollingworth, Hawton, Kapur and Fearn137–Reference Corcoran, Griffin, Arensman, Fitzgerald and Perry139} or attributable to specific factors such as unemployment.^{Reference Mattei and Pistoresi140–Reference Cylus, Glymour and Avendano145} These studies were conducted in Greece,^{Reference Rachiotis, Stuckler, McKee and Hadjichristodoulou135,Reference Alexopoulos, Kavalidou and Messolora136} Italy,^{Reference Mattei and Pistoresi140} Australia,^{Reference Milner, Morrell and LaMontagne141} Spain,^{Reference Ruiz-Perez, Rodriguez-Barranco, Rojas-Garcia and Mendoza-Garcia138,Reference Iglesias-García, Sáiz, Burón, Sánchez-Lasheras, Jiménez-Treviño and Fernández-Artamendi142,Reference Rivera, Casal and Currais143} Barcelona (Spain),^{Reference López-Contreras, Rodríguez-Sanz, Novoa, Borrell, Medallo Muñiz and Gotsens144} the UK,^{Reference Coope, Gunnell, Hollingworth, Hawton, Kapur and Fearn137} Ireland^{Reference Corcoran, Griffin, Arensman, Fitzgerald and Perry139} and the USA.^{Reference Alexopoulos, Kavalidou and Messolora136,Reference Cylus, Glymour and Avendano145} A study from 29 countries in the European Union found a general relationship between the economic environment and suicide rates.^{Reference Fountoulakis, Kawohl, Theodorakis, Kerkhof, Navickas and Höschl146}

A study conducted on the male population of 20 countries in the European Union found job losses to be a determinant of suicide risk, and greater spending on active labour market policies and social capital mitigated risks.^{Reference Reeves, McKee, Gunnell, Chang, Basu and Barr147} A study from 27 European countries, 18 North and South American countries, eight Asian countries, and one African country found that suicide rates increased in the European and North and South American countries, particularly in men and in countries with higher levels of job loss.^{Reference Chang, Stuckler, Yip and Gunnell148} In Italy, periods of economic fluctuations were associated with male suicides, whereas severe economic downturns were associated with increased rates overall,^{Reference Mattei, Pistoresi and De Vogli149} and gross domestic product was associated with suicides because of financial problems.^{Reference Mattei, Ferrari, Pingani and Rigatelli150}

Finally, one study in Piraeus, Greece, found a slight decrease in suicide rates,^{Reference Paraschakis, Michopoulos, Efstathiou, Christodoulou, Boyokas and Douzenis151} and a study including all European Union countries found decreased rates in Austria.^{Reference Stuckler, Basu, Suhrcke, Coutts and McKee127} Also, a study in Crete, Greece, found no overall increase in suicide rates.^{Reference Basta, Vgontzas, Kastanaki, Michalodimitrakis, Kanaki and Koutra152}

Other mental health outcomes

Most of the 25 studies assessing other mental health outcomes (Fig. 3(c)) were conducted on nationally or regionally representative samples, and the clear majority found evidence for increased mental distress.^{Reference Blomqvist, Burström and Backhans153–Reference Gudmundsdottir162} The studies that presented results at the population level included 3479–306 664 participants from Sweden,^{Reference Blomqvist, Burström and Backhans153} the UK,^{Reference Thomson, Niedzwiedz and Katikireddi155} Italy,^{Reference Odone, Landriscina, Amerio and Costa156} Spain,^{Reference Urbanos-Garrido and Lopez-Valcarcel157} England,^{Reference Thomson and Katikireddi154,Reference Katikireddi, Niedzwiedz and Popham159} Australia,^{Reference Parker, Jerrim and Anders160} Iceland,^{Reference Gudmundsdottir162} the Valencian Community in Spain^{Reference Tamayo-Fonseca, Nolasco, Moncho, Barona, Irles and Más158} and 36 mainly European countries.^{Reference Gonza and Burger161} Three of these studies were of high quality^{Reference Thomson and Katikireddi154,Reference Odone, Landriscina, Amerio and Costa156,Reference Gudmundsdottir162} and seven were of low quality.^{Reference Blomqvist, Burström and Backhans153,Reference Thomson, Niedzwiedz and Katikireddi155,Reference Urbanos-Garrido and Lopez-Valcarcel157–Reference Gonza and Burger161} Also, two studies on more defined populations of 2050 medical researchers in Greece,^{Reference Sifaki-Pistolla, Chatzea, Melidoniotis and Mechili163} and 13 000 children aged 4–17 years in the USA,^{Reference Golberstein, Gonzales and Meara164} found decreases in mental health. Both studies were of fair quality.

Some of the population-based studies, ranging from 3755 to 11 743 participants, showed decreases in mental health only among particular population groups,^{Reference Houdmont, Kerr and Addley165–Reference Ruiz-Pérez, Bermúdez-Tamayo and Rodríguez-Barranco171} or under higher rates of precarious employment and lower health spending. These studies were conducted in Spain,^{Reference Bartoll, Palència, Malmusi, Suhrcke and Borrell167,Reference Rajmil, Medina-Bustos, Fernandez de Sanmamed and Mompart-Penina169,Reference Ruiz-Pérez, Bermúdez-Tamayo and Rodríguez-Barranco171} Ireland,^{Reference Houdmont, Kerr and Addley165} Iceland,^{Reference Hauksdóttir, McClure, Jonsson, Olafsson and Valdimarsdóttir166} France^{Reference Malard, Chastang and Niedhammer168} and the UK.^{Reference Lindström and Giordano170} In the USA, retail sales for angiotensin-converting enzyme inhibitors and selective serotonin reuptake inhibitors/serotonin–noradrenaline reuptake inhibitors were not associated with unemployment, but there were positive associations for opioids and phosphodiesterase inhibitors.^{Reference Kozman, Graziul, Gibbons and Alexander172} Five of these studies were of high quality^{Reference Hauksdóttir, McClure, Jonsson, Olafsson and Valdimarsdóttir166,Reference Malard, Chastang and Niedhammer168,Reference Lindström and Giordano170–Reference Kozman, Graziul, Gibbons and Alexander172} and three were or fair quality.^{Reference Houdmont, Kerr and Addley165,Reference Bartoll, Palència, Malmusi, Suhrcke and Borrell167,Reference Rajmil, Medina-Bustos, Fernandez de Sanmamed and Mompart-Penina169}

Also, one study with a cohort of 3321 mothers and 4089 children in Australia found that girls experienced an increase in mental health problems, but not boys or mothers.^{Reference Bubonya, Cobb-Clark, Christensen, Johnson and Zubrick173} This study was of fair quality.

Four studies found no changes in mental health outcomes. They were conducted on a population-based sample in the UK;^{Reference Boyce, Delaney and Wood174} a nationally representative sample of adults aged over 50 years in Ireland;^{Reference Barrett and O'Sullivan175} a study of 21 European countries;^{Reference Sarracino and Piekalkiewicz176} and a study of children aged 11–15 years from Israel, the USA and 31 countries in Europe.^{Reference Rathmann, Pförtner, Hurrelmann, Osorio, Bosakova and Elgar177} One of these studies was of high quality,^{Reference Boyce, Delaney and Wood174} two were of fair quality^{Reference Sarracino and Piekalkiewicz176,Reference Rathmann, Pförtner, Hurrelmann, Osorio, Bosakova and Elgar177} and one was of low quality.^{Reference Barrett and O'Sullivan175}

Healthcare utilisation

Figure 3(d) presents a harvest plot for economic crises and healthcare utilisation. Five of the seven studies assessing changes in healthcare utilisation for mental health problems found increases in rates. They addressed in-patient admissions for affective disorders in Italy,^{Reference Wang and Fattore178} hospital admissions owing to depression in Taiwan,^{Reference Bonnie Lee, Liao and Lin179} primary care patients in Spain,^{Reference Gili, Roca, Basu, McKee and Stuckler180} general practice patients in the UK^{Reference Kendrick, Stuart, Newell, Geraghty and Moore181} and hospital morbidity data in Spain.^{Reference Medel-Herrero and Gomez-Beneyto182} Four studies were of high quality^{Reference Wang and Fattore178,Reference Bonnie Lee, Liao and Lin179,Reference Kendrick, Stuart, Newell, Geraghty and Moore181,Reference Medel-Herrero and Gomez-Beneyto182} and one was of fair quality.^{Reference Gili, Roca, Basu, McKee and Stuckler180}

Two studies did not find overall increases in mental healthcare utilisation: in the UK, rates of self-harm among patients increased in Derby and among males in Manchester, but not in in Oxford;^{Reference Hawton, Bergen, Geulayov, Waters, Ness and Cooper183} in the USA, physician visits owing to mental health disorders decreased after the onset of the crisis, but the use of psychotropic medications increased.^{Reference Chen and Dagher184} Both of these studies were of high quality.