EOL care patients with advanced cancer

In total, 22 studies assessed patients with advanced cancer in EOL care (Supplementary Table 1).^{Reference Al-Shaqi, Alami and Zahrani15–Reference White, Kuhlenschmidt, Vancura and Navari36} All included studies were retrospective cohort studies (18 studies)^{Reference Al-Shaqi, Alami and Zahrani15–Reference Datta, Zhu, Han, Allore, Quagliarello and Juthani-Mehta18,Reference Gao, Spicer and Higginson20–Reference Lin, Lin, Huang, Feng and Tsao25,Reference Mohammed, Al-Zahrani, Sherisher, Alnagar, El-Shentenawy and El-Kashif27–Reference Porta, Rizzi, Lonati, De Luca and Cattaneo32,Reference Schur, Masel, Nemecek, Watzke and Pirker34,Reference Thompson, Silveira, Vitale and Malani35} or prospective cohort studies (4 studies).^{Reference Fombuena Moreno and Espinar Cid19,Reference Mirhosseini, Oneschuk, Hunter, Hanson, Quan and Amigo26,Reference Reinbolt, Shenk, White and Navari33,Reference White, Kuhlenschmidt, Vancura and Navari36} Most of the studies were conducted at an academic medical center (8 studies)^{Reference Azad, Siow, Tafreshi, Moran and Franco16–Reference Datta, Zhu, Han, Allore, Quagliarello and Juthani-Mehta18,Reference Homsi, Walsh, Panta, Lagman, Nelson and Longworth23,Reference Lin, Lin, Huang, Feng and Tsao25,Reference Oh, Kim and Kim28,Reference Pahole, Červek, Žnidarič, Toplak, Zavratnik and Červek30,Reference Thompson, Silveira, Vitale and Malani35} or a hospice or home care (n = 5 studies).^{Reference Girmenia, Moleti and Cartoni21,Reference Helde-Frankling, Bergqvist, Bergman and Björkhem-Bergman22,Reference Oneschuk, Fainsinger and Demoissac29,Reference Porta, Rizzi, Lonati, De Luca and Cattaneo32,Reference White, Kuhlenschmidt, Vancura and Navari36} EOL care varied from within 1 day prior to death in 1 study^{Reference Baek, Chang, Byun, Han and Heo17} to <6 months in another study.^{Reference White, Kuhlenschmidt, Vancura and Navari36} In 14 EOL studies, the time before death was not stated.^{Reference Azad, Siow, Tafreshi, Moran and Franco16,Reference Datta, Zhu, Han, Allore, Quagliarello and Juthani-Mehta18,Reference Fombuena Moreno and Espinar Cid19,Reference Girmenia, Moleti and Cartoni21,Reference Homsi, Walsh, Panta, Lagman, Nelson and Longworth23,Reference Lin, Lin, Huang, Feng and Tsao25–Reference Mohammed, Al-Zahrani, Sherisher, Alnagar, El-Shentenawy and El-Kashif27,Reference Oneschuk, Fainsinger and Demoissac29,Reference Pereira, Watanabe and Wolch31–Reference Thompson, Silveira, Vitale and Malani35}

In combination, these studies included 35,239 patients with advanced cancer. Moreover, 16 studies (4,707 patients, 13.4%) included patients with any type of cancer (solid or hematologic malignancy).^{Reference Al-Shaqi, Alami and Zahrani15–Reference Fombuena Moreno and Espinar Cid19,Reference Helde-Frankling, Bergqvist, Bergman and Björkhem-Bergman22–Reference Lin, Lin, Huang, Feng and Tsao25,Reference Mohammed, Al-Zahrani, Sherisher, Alnagar, El-Shentenawy and El-Kashif27–Reference Oneschuk, Fainsinger and Demoissac29,Reference Pereira, Watanabe and Wolch31,Reference Reinbolt, Shenk, White and Navari33,Reference Thompson, Silveira, Vitale and Malani35,Reference White, Kuhlenschmidt, Vancura and Navari36} In 17 studies in which data were available, fewer than one-sixth of patients (4,332, 12.3%) with advanced cancer were referred to palliative care.^{Reference Datta, Zhu, Han, Allore, Quagliarello and Juthani-Mehta18,Reference Fombuena Moreno and Espinar Cid19,Reference Girmenia, Moleti and Cartoni21–Reference Mohammed, Al-Zahrani, Sherisher, Alnagar, El-Shentenawy and El-Kashif27,Reference Oneschuk, Fainsinger and Demoissac29–Reference White, Kuhlenschmidt, Vancura and Navari36} In 15 studies (68%), >50% of patients received antibiotics during their EOL period.^{Reference Baek, Chang, Byun, Han and Heo17–Reference Fombuena Moreno and Espinar Cid19,Reference Girmenia, Moleti and Cartoni21,Reference Homsi, Walsh, Panta, Lagman, Nelson and Longworth23–Reference Oh, Kim and Kim28,Reference Pahole, Červek, Žnidarič, Toplak, Zavratnik and Červek30,Reference Pereira, Watanabe and Wolch31,Reference Reinbolt, Shenk, White and Navari33,Reference Thompson, Silveira, Vitale and Malani35,Reference White, Kuhlenschmidt, Vancura and Navari36} Furthermore, 7 studies did not report the antibiotic route of administration,^{Reference Azad, Siow, Tafreshi, Moran and Franco16,Reference Baek, Chang, Byun, Han and Heo17,Reference Gao, Spicer and Higginson20,Reference Oh, Kim and Kim28,Reference Pahole, Červek, Žnidarič, Toplak, Zavratnik and Červek30,Reference Reinbolt, Shenk, White and Navari33,Reference Schur, Masel, Nemecek, Watzke and Pirker34} and 16 studies did not report the duration of antibiotic therapy.^{Reference Al-Shaqi, Alami and Zahrani15–Reference Baek, Chang, Byun, Han and Heo17,Reference Gao, Spicer and Higginson20–Reference Helde-Frankling, Bergqvist, Bergman and Björkhem-Bergman22,Reference Lam, Chan, Tse and Leung24,Reference Lin, Lin, Huang, Feng and Tsao25,Reference Mohammed, Al-Zahrani, Sherisher, Alnagar, El-Shentenawy and El-Kashif27,Reference Oneschuk, Fainsinger and Demoissac29–Reference Schur, Masel, Nemecek, Watzke and Pirker34,Reference White, Kuhlenschmidt, Vancura and Navari36} None of these studies reported the antibiotic consumption in days of therapy per 1,000 patient days. Microbiological cultures were requested before antibiotics were prescribed in 15 studies,^{Reference Al-Shaqi, Alami and Zahrani15,Reference Datta, Zhu, Han, Allore, Quagliarello and Juthani-Mehta18,Reference Fombuena Moreno and Espinar Cid19,Reference Girmenia, Moleti and Cartoni21–Reference Oh, Kim and Kim28,Reference Pereira, Watanabe and Wolch31,Reference Reinbolt, Shenk, White and Navari33,Reference Thompson, Silveira, Vitale and Malani35,Reference White, Kuhlenschmidt, Vancura and Navari36} and 7 studies did not report microbiological data.^{Reference Azad, Siow, Tafreshi, Moran and Franco16,Reference Baek, Chang, Byun, Han and Heo17,Reference Gao, Spicer and Higginson20,Reference Oneschuk, Fainsinger and Demoissac29,Reference Pahole, Červek, Žnidarič, Toplak, Zavratnik and Červek30,Reference Porta, Rizzi, Lonati, De Luca and Cattaneo32,Reference Schur, Masel, Nemecek, Watzke and Pirker34} Only 12 studies examined whether the use of antibiotics was associated with beneficial outcomes (survival or comfort),^{Reference Datta, Zhu, Han, Allore, Quagliarello and Juthani-Mehta18,Reference Fombuena Moreno and Espinar Cid19,Reference Girmenia, Moleti and Cartoni21,Reference Helde-Frankling, Bergqvist, Bergman and Björkhem-Bergman22,Reference Lam, Chan, Tse and Leung24–Reference Oh, Kim and Kim28,Reference Porta, Rizzi, Lonati, De Luca and Cattaneo32,Reference Reinbolt, Shenk, White and Navari33,Reference White, Kuhlenschmidt, Vancura and Navari36} and 1 of these evaluated potential adverse effects associated with antibiotic use.^{Reference Reinbolt, Shenk, White and Navari33} These studies had high risk of bias, with scores (up to 28 points) ranging from 4 to 20 with a median of 12.

EOL patients with advanced dementia

Overall, 17 studies assessed patients with advanced dementia in EOL care (Supplementary Table 2).^{Reference Ahronheim, Morrison, Morris, Baskin and Meier37–Reference Volicer, Hurley, Fabiszewski, Montgomery and Volicer53} Studies were conducted in nursing homes (6 studies)^{Reference D’Agata and Mitchell40,Reference D’Agata, Loeb and Mitchell41,Reference Givens, Jones, Shaffer, Kiely and Mitchell45,Reference Mitchell, Shaffer and Loeb49,Reference van der Steen, Lane, Kowall, Knol and Volicer51,Reference van der Steen, Di Giulio and Giunco52} and long-term care facilities (3 studies).^{Reference Chen, Lamberg and Chen39,Reference Di Giulio, Toscani, Villani, Brunelli, Gentile and Spadin42,Reference Klapwijk, Caljouw, van Soest-Poortvliet, van der Steen and Achterberg47} The EOL care period evaluated varied from the 48 hours prior to death in 1 study^{Reference Hirakawa, Masuda, Kuzuya, Kimata, Iguchi and Uemura46} to 6 months in 2 studies.^{Reference Chen, Lamberg and Chen39,Reference Volicer, Hurley, Fabiszewski, Montgomery and Volicer53} In 7 EOL studies, the time before death was not stated.^{Reference Ahronheim, Morrison, Morris, Baskin and Meier37,Reference Catic, Berg and Moran38,Reference D’Agata, Loeb and Mitchell41,Reference Evers, Purohit, Perl, Khan and Marin43–Reference Givens, Jones, Shaffer, Kiely and Mitchell45,Reference Marttini Abarca, Carrillo Garcia, Oviedo Briones and Perdomo Ramirez48}

Most studies utilized a prospective study design (9 studies)^{Reference D’Agata and Mitchell40,Reference D’Agata, Loeb and Mitchell41,Reference Fabiszewski, Volicer and Volicer44,Reference Givens, Jones, Shaffer, Kiely and Mitchell45,Reference Klapwijk, Caljouw, van Soest-Poortvliet, van der Steen and Achterberg47,Reference Nourhashémi, Gillette and Cantet50–Reference Volicer, Hurley, Fabiszewski, Montgomery and Volicer53} or retrospective cohort study design (7 studies).^{Reference Catic, Berg and Moran38,Reference Chen, Lamberg and Chen39,Reference Di Giulio, Toscani, Villani, Brunelli, Gentile and Spadin42,Reference Evers, Purohit, Perl, Khan and Marin43,Reference Hirakawa, Masuda, Kuzuya, Kimata, Iguchi and Uemura46,Reference Marttini Abarca, Carrillo Garcia, Oviedo Briones and Perdomo Ramirez48,Reference Mitchell, Shaffer and Loeb49} One study was a randomized clinical trial.^{Reference Ahronheim, Morrison, Morris, Baskin and Meier37} In combination, these studies included 2,501 patients with advanced dementia. In the 5 studies (29.7%) in which data were available, fewer than one-quarter of patients (498, 19.9%) with advanced dementia were referred to palliative care.^{Reference Ahronheim, Morrison, Morris, Baskin and Meier37,Reference Catic, Berg and Moran38,Reference Fabiszewski, Volicer and Volicer44,Reference Mitchell, Shaffer and Loeb49,Reference Volicer, Hurley, Fabiszewski, Montgomery and Volicer53} In 12 studies (71%), >50% of patients received antibiotics during the EOL period,^{Reference Ahronheim, Morrison, Morris, Baskin and Meier37–Reference Chen, Lamberg and Chen39,Reference D’Agata, Loeb and Mitchell41–Reference Givens, Jones, Shaffer, Kiely and Mitchell45,Reference Mitchell, Shaffer and Loeb49,Reference van der Steen, Lane, Kowall, Knol and Volicer51–Reference Volicer, Hurley, Fabiszewski, Montgomery and Volicer53} and 10 studies did not report the antibiotic route of administration.^{Reference D’Agata and Mitchell40–Reference Evers, Purohit, Perl, Khan and Marin43,Reference Hirakawa, Masuda, Kuzuya, Kimata, Iguchi and Uemura46–Reference Marttini Abarca, Carrillo Garcia, Oviedo Briones and Perdomo Ramirez48,Reference Nourhashémi, Gillette and Cantet50,Reference van der Steen, Di Giulio and Giunco52,Reference Volicer, Hurley, Fabiszewski, Montgomery and Volicer53} Furthermore, 15 studies did not report the duration of antibiotic therapy,^{Reference Ahronheim, Morrison, Morris, Baskin and Meier37–Reference Marttini Abarca, Carrillo Garcia, Oviedo Briones and Perdomo Ramirez48,Reference Nourhashémi, Gillette and Cantet50,Reference van der Steen, Di Giulio and Giunco52,Reference Volicer, Hurley, Fabiszewski, Montgomery and Volicer53} and only 2 studies reported the antibiotic consumption in days-of-therapy (DOT) per 1,000 resident days.^{Reference D’Agata and Mitchell40,Reference Mitchell, Shaffer and Loeb49} Microbiological cultures were requested before the antibiotic had been prescribed in 2 studies,^{Reference Fabiszewski, Volicer and Volicer44,Reference Mitchell, Shaffer and Loeb49} and 15 studies did not report microbiological data.^{Reference Ahronheim, Morrison, Morris, Baskin and Meier37–Reference Evers, Purohit, Perl, Khan and Marin43,Reference Givens, Jones, Shaffer, Kiely and Mitchell45–Reference Marttini Abarca, Carrillo Garcia, Oviedo Briones and Perdomo Ramirez48,Reference Nourhashémi, Gillette and Cantet50–Reference Volicer, Hurley, Fabiszewski, Montgomery and Volicer53} Only 3 studies examined whether the use of antibiotics was associated with beneficial outcomes (ie, comfort),^{Reference Mitchell, Shaffer and Loeb49,Reference van der Steen, Di Giulio and Giunco52,Reference Volicer, Hurley, Fabiszewski, Montgomery and Volicer53} and only 3 studies evaluated a survival benefit of antibiotic exposure.^{Reference Fabiszewski, Volicer and Volicer44,Reference Givens, Jones, Shaffer, Kiely and Mitchell45,Reference van der Steen, Lane, Kowall, Knol and Volicer51} None of these studies evaluated the potential adverse effects associated with antibiotic use.^{Reference Ahronheim, Morrison, Morris, Baskin and Meier37–Reference Volicer, Hurley, Fabiszewski, Montgomery and Volicer53} Studies had a moderate risk of bias, with scores (up to 28 points) ranging from 7 to 22 with a median of 18.

EOL patients in “mixed populations”

Overall, 33 studies documented patients in EOL care classified as a “mixed population” (Supplementary Table 3).^{Reference Ahronheim, Morrison, Baskin, Morris and Meier54–Reference Vitetta, Kenner and Sali86} Most studies were conducted at an academic medical center (19 studies)^{Reference Ahronheim, Morrison, Baskin, Morris and Meier54,Reference Alonso, Ebert, Dörr, Buchheidt, Hennerici and Szabo56,Reference Arribas, Serrano and Herraiz57,Reference Burnham, Chi, Ma, Dans and Kollef60–Reference Chun, Rodgers, Vitale, Collins and Malani62,Reference Dagli, Tasdemir and Ulutasdemir64,Reference Fins, Miller, Acres, Bacchetta, Huzzard and Rapkin65,Reference Furuno, Noble and Horne67,Reference Kadoyama, Noble and Izumi70,Reference Merel, Meier, McKinney and Pottinger73–Reference Phua, Kee and Tan77,Reference Robinson, Ravilly, Berde and Wohl80,Reference Smallwood, Bartlett, Taverner, Le, Irving and Philip81,Reference Taverner, Ross, Bartlett, Luthe, Ong, Irving and Smallwood84,Reference Vitetta, Kenner and Sali86} or a hospice, home care, or nursing home (9 studies).^{Reference Albrecht, McGregor, Fromme, Bearden and Furuno55,Reference Brabin and Allsopp59,Reference Fong, Oravec and Radwany66,Reference Furuno, Noble, Bearden and Fromme68,Reference Koon and Heng71,Reference Reinhardt, Downes, Cimarolli and Bomba79,Reference Soh, Van Der Poel, Mcquillan and Kennely82,Reference Tiirola, Korhonen, Surakka and Lehto85} The EOL care period definition varied from the 12 hours prior to death in 1 study^{Reference Robinson, Ravilly, Berde and Wohl80} to 6 months in 2 studies.^{Reference Rajala, Lehto, Saarinen, Sutinen, Saarto and Myllärniemi78,Reference Reinhardt, Downes, Cimarolli and Bomba79} In 20 EOL studies, the time before death was not stated.^{Reference Ahronheim, Morrison, Baskin, Morris and Meier54,Reference Alonso, Ebert, Dörr, Buchheidt, Hennerici and Szabo56–Reference Dagli, Tasdemir and Ulutasdemir64,Reference Furuno, Noble and Horne67,Reference Kadoyama, Noble and Izumi70,Reference Merel, Meier, McKinney and Pottinger73–Reference Niederman75,Reference Phua, Kee and Tan77,Reference Smallwood, Bartlett, Taverner, Le, Irving and Philip81,Reference Taverner, Ross, Bartlett, Luthe, Ong, Irving and Smallwood84,Reference Vitetta, Kenner and Sali86}

Most of the included studies were retrospective studies (27 studies),^{Reference Ahronheim, Morrison, Baskin, Morris and Meier54–Reference Alonso, Ebert, Dörr, Buchheidt, Hennerici and Szabo56,Reference Bauduer, Capdupuy and Renoux58,Reference Brabin and Allsopp59,Reference Choi, Kim, Kim and Kim61,Reference Chun, Rodgers, Vitale, Collins and Malani62,Reference Dagli, Tasdemir and Ulutasdemir64–Reference Fong, Oravec and Radwany66,Reference Furuno, Noble, Bearden and Fromme68–Reference Rajala, Lehto, Saarinen, Sutinen, Saarto and Myllärniemi78,Reference Robinson, Ravilly, Berde and Wohl80,Reference Smallwood, Bartlett, Taverner, Le, Irving and Philip81,Reference Tagashira, Kawahara, Takamatsu and Honda83–Reference Vitetta, Kenner and Sali86} prospective cohort studies (3 studies),^{Reference Arribas, Serrano and Herraiz57,Reference Clayton, Fardell, Hutton-Potts, Webb and Chye63,Reference Soh, Van Der Poel, Mcquillan and Kennely82} or cross-sectional studies (2 studies),^{Reference Furuno, Noble and Horne67,Reference Reinhardt, Downes, Cimarolli and Bomba79} and 1 study was a randomized clinical trial.^{Reference Burnham, Chi, Ma, Dans and Kollef60} In combination, these studies included 13,028 patients with different terminal illnesses. Also, 11 studies were restricted to cancer and other terminal illness (eg, COPD)^{Reference Bauduer, Capdupuy and Renoux58,Reference Brabin and Allsopp59,Reference Clayton, Fardell, Hutton-Potts, Webb and Chye63,Reference Fins, Miller, Acres, Bacchetta, Huzzard and Rapkin65,Reference Kadoyama, Noble and Izumi70,Reference Koon and Heng71,Reference Merel, Meier, McKinney and Pottinger73–Reference Niederman75,Reference Rajala, Lehto, Saarinen, Sutinen, Saarto and Myllärniemi78,Reference Vitetta, Kenner and Sali86} and 10 studies were restricted to patients with cancer or advanced dementia or with other terminal illness (eg, heart failure or chronic renal failure).^{Reference Ahronheim, Morrison, Baskin, Morris and Meier54,Reference Albrecht, McGregor, Fromme, Bearden and Furuno55,Reference Arribas, Serrano and Herraiz57,Reference Chun, Rodgers, Vitale, Collins and Malani62,Reference Dagli, Tasdemir and Ulutasdemir64,Reference Furuno, Noble and Horne67,Reference Hirakawa, Masuda, Kuzuya, Iguchi, Asahi and Uemura69,Reference Phua, Kee and Tan77,Reference Reinhardt, Downes, Cimarolli and Bomba79,Reference Tagashira, Kawahara, Takamatsu and Honda83} Other studies included COPD patients (2 studies),^{Reference Smallwood, Bartlett, Taverner, Le, Irving and Philip81,Reference Taverner, Ross, Bartlett, Luthe, Ong, Irving and Smallwood84} cystic fibrosis patients (2 studies),^{Reference Philip, Gold and Sutherland76,Reference Robinson, Ravilly, Berde and Wohl80} patients with stroke (1 study),^{Reference Alonso, Ebert, Dörr, Buchheidt, Hennerici and Szabo56} patients with amyotrophic lateral sclerosis (1 study),^{Reference Tiirola, Korhonen, Surakka and Lehto85} elderly patients (1 study),^{Reference Koon and Heng71} elderly patients with and without dementia (1 study),^{Reference Soh, Van Der Poel, Mcquillan and Kennely82} and 4 studies did not report underlying conditions.^{Reference Burnham, Chi, Ma, Dans and Kollef60,Reference Choi, Kim, Kim and Kim61,Reference Fong, Oravec and Radwany66,Reference Furuno, Noble, Bearden and Fromme68}

In 24 studies for which data were available (72.7%), more than a half of patients (58.4%, 7,607) in the “mixed population” were referred to palliative care.^{Reference Alonso, Ebert, Dörr, Buchheidt, Hennerici and Szabo56,Reference Brabin and Allsopp59–Reference Koon and Heng71,Reference Merel, Meier, McKinney and Pottinger73–Reference Philip, Gold and Sutherland76,Reference Robinson, Ravilly, Berde and Wohl80–Reference Vitetta, Kenner and Sali86} In 21 studies, >50% of patients received antibiotics during their EOL period.^{Reference Ahronheim, Morrison, Baskin, Morris and Meier54,Reference Alonso, Ebert, Dörr, Buchheidt, Hennerici and Szabo56,Reference Arribas, Serrano and Herraiz57,Reference Brabin and Allsopp59–Reference Dagli, Tasdemir and Ulutasdemir64,Reference Low, Yap, Chan and Tang72–Reference Philip, Gold and Sutherland76} Also, 23 studies did not report the antibiotic route of administration,^{Reference Albrecht, McGregor, Fromme, Bearden and Furuno55–Reference Bauduer, Capdupuy and Renoux58,Reference Burnham, Chi, Ma, Dans and Kollef60–Reference Chun, Rodgers, Vitale, Collins and Malani62,Reference Dagli, Tasdemir and Ulutasdemir64,Reference Fins, Miller, Acres, Bacchetta, Huzzard and Rapkin65,Reference Furuno, Noble and Horne67–Reference Low, Yap, Chan and Tang72,Reference Niederman and Ornstein74–Reference Reinhardt, Downes, Cimarolli and Bomba79,Reference Smallwood, Bartlett, Taverner, Le, Irving and Philip81,Reference Tiirola, Korhonen, Surakka and Lehto85} and 26 studies did not report the duration of antibiotic therapy.^{Reference Ahronheim, Morrison, Baskin, Morris and Meier54–Reference Alonso, Ebert, Dörr, Buchheidt, Hennerici and Szabo56,Reference Bauduer, Capdupuy and Renoux58,Reference Brabin and Allsopp59,Reference Choi, Kim, Kim and Kim61–Reference Clayton, Fardell, Hutton-Potts, Webb and Chye63,Reference Fong, Oravec and Radwany66–Reference Kadoyama, Noble and Izumi70,Reference Low, Yap, Chan and Tang72–Reference Soh, Van Der Poel, Mcquillan and Kennely82,Reference Tiirola, Korhonen, Surakka and Lehto85,Reference Vitetta, Kenner and Sali86} Only 1 study reported the antibiotic consumption in days of therapy (DOT) per 1,000 patient days.^{Reference Tagashira, Kawahara, Takamatsu and Honda83} Microbiological cultures were requested before antibiotics were prescribed in 8 studies,^{Reference Brabin and Allsopp59,Reference Chun, Rodgers, Vitale, Collins and Malani62–Reference Dagli, Tasdemir and Ulutasdemir64,Reference Furuno, Noble and Horne67,Reference Tagashira, Kawahara, Takamatsu and Honda83,Reference Taverner, Ross, Bartlett, Luthe, Ong, Irving and Smallwood84,Reference Vitetta, Kenner and Sali86} and 25 of these studies did not report microbiological data.^{Reference Ahronheim, Morrison, Baskin, Morris and Meier54–Reference Bauduer, Capdupuy and Renoux58,Reference Burnham, Chi, Ma, Dans and Kollef60,Reference Choi, Kim, Kim and Kim61,Reference Fins, Miller, Acres, Bacchetta, Huzzard and Rapkin65,Reference Fong, Oravec and Radwany66,Reference Furuno, Noble, Bearden and Fromme68–Reference Soh, Van Der Poel, Mcquillan and Kennely82,Reference Tiirola, Korhonen, Surakka and Lehto85} Only 6 studies examined whether the use of antibiotics was associated with beneficial outcomes (ie, comfort),^{Reference Clayton, Fardell, Hutton-Potts, Webb and Chye63,Reference Fins, Miller, Acres, Bacchetta, Huzzard and Rapkin65,Reference Koon and Heng71,Reference Merel, Meier, McKinney and Pottinger73,Reference Tagashira, Kawahara, Takamatsu and Honda83,Reference Vitetta, Kenner and Sali86} and 1 study evaluated survival and comfort as beneficial outcome due to antibiotic prescription.^{Reference Brabin and Allsopp59} Most of these studies (26 studies) did not report any outcome associated with antibiotic prescriptions. These studies had high risk of bias, with scores (up to 28 points) ranging from 6 to 19, with a median of 13.

Palliative care consultation and antibiotic use measures

When the results of the studies were pooled, there was a statistically significant association between receipt of palliative care consultation and higher odds of receiving antibiotics (3 studies; pooled odds ratio, 1.73; 95% CI, 1.02–2.93) (Fig. 2).^{Reference Ahronheim, Morrison, Morris, Baskin and Meier37,Reference Catic, Berg and Moran38,Reference Niederman and Ornstein74} The result of this meta-analysis was homogeneous (heterogeneity P = .94; I^{Reference Thomas and Armstrong-Wilson2} = 0%) (Fig. 2). Two studies were of patients with advanced dementia,^{Reference Ahronheim, Morrison, Morris, Baskin and Meier37,Reference Catic, Berg and Moran38} and 1 study included a “mixed population” patients.^{Reference Niederman and Ornstein74} Notably, we were unable to include other studies evaluating palliative care consultation in the meta-analysis because the necessary data were not included in the published studies. There was no evidence for publication bias among these studies (P = .40).

Fig. 2. Forest plot of patients receiving antibiotics (n = 3 studies) under palliative care consultation (PCC) and nonpalliative care consultation (non-PCC). Odds ratios (ORs) were determined with the Mantel-Haenszel random-effects method. Note. CI, confidence interval; M-H, Mantel-Haenszel.