Overview of articles
Twenty-six articles were read in full text of which nineteen articles were excluded because they did not meet the inclusion criteria, for instance, they did not focus on art making or did not include an artist or art therapist. Hence, seven articles were suitable for further analysis. These included three non-randomized intervention studies and four randomized controlled trials. An overview of our selection strategy can be found in Fig. 1.
The number of participants reported in the articles varied between 24 and 183. Six out of seven papers included more than 50 patients. Three studies focused on female cancer patients only [22,23,24]. In the remaining articles, patents with a variety of cancer diagnoses were included. In general, more women than men participated in the art therapy trials. A complete overview of the sociodemographic characteristics was given in all studies, except for Radl et al.  who only reported age and race of the participants. Four articles described the diagnosis of the patients and their clinical characteristics [24,25,26,27].
All studies used a quantitative design with validated outcome measures. The Hospital Anxiety and Depression Scale (HADS) and the EORTC-QLQ C-30 were used most frequently as outcome measures. One study added a qualitative questionnaire to explore the satisfaction with the art therapy intervention .
The non-randomized controlled trials were assessed based on nine questions about the methodological quality of the studies (Table 1). In all articles, the examined causes and effects were clear. The measurements were psychometrically robust and were applied both before and after the interventions. However, the patients in the control group were only similar to the patients in the intervention group in one study . For example, in one study, the control group consisted of patients who declined participation in the art therapy program, which may have caused selection bias . Also, it was often unclear whether the control group and the intervention group received similar cancer treatment apart from the art therapy intervention [25, 27].
The checklist regarding randomized controlled trials consisted of twelve questions. In all studies, true randomization was used; however, blinding the treatment was self-evidently not applicable in any of the studies [22, 23]. Porter and McConnell  noted that their outcome assessors were blinded. All seven articles used appropriate statistical analysis. Full elaboration of the answers to the questions on the checklists can be found in Supplementary Material. No studies were excluded based on their methodological quality.
Description of the included articles
Bozcuk et al. 
Bozcuk and Ozcan  included participants from the outpatient chemotherapy unit Akdeniz University Medical Faculty in Antalya, Turkey. Patients were classified based on their previous exposure to painting art therapy and were divided into two intervention groups. Patients declining participation served as control group. An art therapist with experience in painting art therapy worked with everyone individually. First, he provided information about the materials and techniques and then let the patients make as much watercolor paintings as they wanted during a chemotherapy appointment. Afterward, the art therapist encouraged the patients to elaborate on the meaning and subject of their finished work. The number of finished paintings was registered as a representation of motivation.
De Feudis et al. 
De Feudis and Graziano  provided art therapy sessions of 90 min in the Medical Oncology Out-Patient unit of San Paolo, serving a population of adult cancer patients from Puglia, Italy. Each patient participated in one group session. The control group was on a waiting list to receive art therapy and meanwhile received usual care. A psychotherapist skilled in art therapy guided the sessions, assisted by a psycho-oncology team. The intervention took place in a room equipped with a large amount of art materials and background music. Groups consisted of a maximum of eight people, varying in age, gender, and diagnosis. The therapy focused on three principles: production of spontaneous artwork, provocation of self-reflection, and sharing experiences with group members. Afterward, all patients were offered the opportunity of additional psychosocial support.
Geue et al. 
The hemato-oncological patients in the study of Geue and Richter  were recruited from the Leipzig University Hospital, Germany. Hemato-oncological patients who lived too far away to participate formed the control group. Twenty-two weekly sessions of 90 min were held under the supervision of an art therapist. The groups included patients of different gender and age. The intervention consisted of three phases: becoming familiar with drawing, assisted by an artist, watercolor painting by oneself, and creating an individual book to express feelings. All decisions regarding the content or design of the book were made by patients themselves.
Jalambadani et al. 
Jalambadani and Borji  investigated Neyshabur women with breast cancer visiting the Razavi Hospital of Mashhad City, Iran. They conducted twelve weekly mindfulness-based art therapy (MBAT) sessions, lasting on average 90 min. The control group was on a waiting list to receive art therapy and was provided with usual cancer care. The MBAT-program focused on the procedure first used in Monti and Peterson , involving an introduction to art-making, self-picture assessment tasks, exploration of art materials and mind-body relationship, creative problem-solving, meditation, free art-making, and group discussions. The interventions were guided by an artist with psycho-oncological training.
Jang et al. 
Jang and Kang  examined the effects of mindfulness-based art therapy (MBAT) in women with breast cancer, who had received surgery and radiation therapy at Wonkwang University Hospital, South-Korea. The patients in the MBAT-group were provided with twelve weekly sessions lasting 45 min each. The qualified art therapist encourages the patients to express their inner feelings. Both the intervention group and the control group continued to have standard post-treatment care.
Porter et al. 
Porter, McConnell  developed music therapy sessions for hospice patients in Northern Ireland with an Eastern Cooperative Oncology Group (ECOG) performance of 2 or lower. The intervention group received a total of six 45-min individual music therapy sessions, twice a week. The control group underwent usual cancer care. A trained and registered music therapist provided the program using an interactive approach. Patients could participate by singing or listening to known music, but they also got the opportunity to create something of their own, e.g., a melody, song, rhythm, or instrumental piece. The music therapist supported the patients in the creative process.
Radl et al. 
Self-Book art therapy was offered by Radl and Vita  to female cancer patients undergoing active oncological treatment in a major hospital in Philadelphia, USA. Both the intervention group and the control group had access to all available complementary (psychological) therapies, but only the intervention group created a Self-Book. The participants worked with an art therapist individually in six sessions (“agreements”) of about 50 min. The purpose of the art therapy was to create a self-reflective book to express one’s feelings and experiences. During the first five sessions, the patients were instructed to fill the pages of their book with creative artwork related to a given subject (safe place, supports, strength and virtues, wishes for loved ones, wishes for oneself). In the final session, the patients were encouraged to decorate the cover of the book.
Findings: effect on outcome measures
Out of the four studies measuring anxiety, two found a significant improvement. De Feudis and Graziano  reported a significant reduction in anxiety scores in the intervention group, with the score decreasing from 44.3 to 37.1 (p = 0.002), while the anxiety scores in the control group did not significantly change. However, the study did not find a significant difference in anxiety scores between the two groups. In the study of Jang and Kang , anxiety scores were significantly improved compared to the control group (p < 0.001). Geue and Richter  did not find any significant differences, neither within the invention group nor between the intervention group and control group. Bozcuk and Ozcan  compared anxiety scores among two intervention groups and one control group and found that anxiety scores did not differ significantly between the group.
Three of the seven studies compared depression scores between intervention and control groups. Jang and Kang  and Bozcuk and Ozcan  found the depression scores in the intervention group to significantly improved compared to the control group (p < 0.001 and p = 0.001 respectively). Geue and Richter  found neither significant improvement in depression scores within the groups nor between the groups.
Quality of life
Six studies reported on QoL or QoL related scales, such as well-being, of which four found an improvement in these outcome variables. Bozcuk and Ozcan  reported a significant difference in QoL between the intervention groups and the control group (p = 0.001). In addition, as expected through the regression to the mean principle, patients with lower QoL appeared to take the greatest advantage from painting art therapy program. All participants declared they enjoyed taking part in painting art therapy program. The intervention was also found to be feasible during chemotherapy sessions. Jang and Kang  also reported improvement in quality of life, with the global health status/QoL score increasing from 26.4 to 81.3 (p < 0.001). Significant beneficial effects on functional scales, physical symptoms, and financial difficulties were also noted. None of these changes were found in the control group. Additionally, Jalambadani and Borji  showed statistically significant decreases in symptoms of distress in the intervention group compared to the waiting list control group. The scores of physical health, psychological symptoms, social relationships, and environmental factors were improved significantly, as well as quality of life behavior. Lastly, De Feudis and Graziano  reported that 89.3% of the participating patients considered the art therapy program beneficial to their well-being.
Radl and Vita  documented no statistically significant differences between the Self-Book therapy intervention group and the control group for the primary outcome (emotional distress) or the secondary outcome (psychological well-being). However, they did find significant improvement in the spiritual well-being of the patients taking part in the Self-Book art therapy program. Also in the study of Porter and McConnell  changes in McGill Quality of Life questionnaire (MQoL) scores, as well as in physical symptoms and psychological and existential well-being, from baseline to the first assessment (week 1) were not statistically different between the intervention group and the control group .
In conclusion, of the seven studies, four identified significant results regarding anxiety, depression, or QoL [22, 23, 25, 26]. Of the four studies that studied anxiety, half found significant improvements in anxiety scores, the other half did not [23, 26]. Regarding depression, two studies found significant improvement in depression scores and one did not [23, 25]. Four out of six studies regarding QoL showed significant improvement in QoL after the art therapy intervention [22, 23, 25]. Hence, three studies did not identify any significant results regarding anxiety, depression, or QoL [24, 27, 28]. Nevertheless, all participants considered the experience valuable to their well-being, what came up anecdotally as well as through questionnaires after completion of the intervention. An overview of the results of all studies can be found in Table 2.