Patient sample and data
Data was retrieved from the prospective controlled randomised intervention trial Tecla. The inclusion criteria of Tecla were a medical diagnosis of any form of schizophrenia (ICD-10 F20), schizoaffective disorders (ICD-10 F25) or bipolar disorders (ICD-10 F31) and an age ≥ 18 years. The diagnoses were extracted from the patient files. The exclusion criteria were prior scheduled inpatient treatment within the next six months and lack of ability to be contacted by mobile phone. The participants were recruited shortly before their discharge from day-care hospitals or open or locked inpatient wards from three psychiatric departments in Western-Pomerania, a federal state in the North East part of Germany.
Tecla has been approved by the Ethics Committee of the University Medicine Greifswald (BB 122/14) and was registered on the German Clinical Trials Register (date 2015\05\21, DRKS00008548). A comprehensive description of the study protocol for the Tecla study was published by Stentzel et al. [14].
Randomisation
The participants were randomized in the intervention or control group after the baseline assessment. A scientist who was neither involved in the recruitment nor in the baseline assessment performed the allocation to the groups using a random allocation (block randomisation) on a blind basis. The listing of the two groups was irregular. The participants with chronic conditions were assigned to the next entry in the randomisation list.
Telemedical intervention
Participants were individually randomised in the intervention group and control group. Both groups received care as usual in the outpatient facilities (outpatient psychiatric / psychotherapeutic practices or psychiatric institutional outpatient departments). The intervention group received regular telephone calls every two weeks and standardised and individualised text messages every week. An example of an individualised text message is provided in Figure 1. Qualified nurses who specialise in telemedical care conducted the regular telephone calls. The nurses are part of regular meetings at one of the psychiatric institutions’ outpatient departments and day-care hospital. They received training in how to use the documentation system and have completed appropriate psychiatric/psychotherapeutic education programmes. The telemedical conversation was conducted on the basis of eCRFs in a computer-aided documentation system in accordance with the current standards for data security and data privacy [15, 16]. The standardised conversation contained a structured standardised part and an individualised part. The structured standardised part of the telephone calls included suicidal tendencies, changes in the medication regime, medication adherence and medication side effects (study protocol published elsewhere [13]). The individualised part addressed selected topics of everyday life that the respective participant evaluated as important for themselves and their condition. The weekly text messages refer to actual and relevant events and themes in the daily life of the participants.
Measures
WHOQOL-BREF
Quality of life was measured with WHOQOL-BREF, the short version of the World Health Organization Quality of Life subjective instrument, which is designed for generic use [9, 17]. It assesses quality of life from a subjective perspective [7]. The short version, WHOQOL-BREF, has 26 items. Answers are given on 1-to-5-point Likert scales. The higher the total score, the better the patient’s quality of life [17]. WHOQOL assesses different aspects of life that are relevant for quality of life [9]. The WHOQOL-BREF is based on four domains [9, 17] and one global value for general quality of life:
- Physical domain: pain, energy, sleep, mobility, activities, medication, work.
- Psychological domain: positive feelings, cognitions, self-esteem, body image, negative feelings, spirituality.
- Social relationships: personal relationships, social support, sex.
- Environment: safety and security, home environment, finance, health/social care, information, leisure, physical environment, transport.
- Global: overall quality of life, general health.
The German version was used, which shows good internal consistence (Cronbach’s α > 0.7 for all domains) for the overall population as well as for patients with mental illnesses [18].
Participants’ evaluation of the telemedical care programme
Participants in the intervention group were asked to evaluate the telemedical care at the end of their time participating in the study by answering the questions shown in Table 1.
Table 1: Interview questions and answers to assess the participants’ acceptance and satisfaction
Question:
|
How would you describe the telephone and text message contact during the last 6 months?
|
Answer:
|
Very helpful – moderately helpful – not helpful – other (free text) – don’t know – no answer
|
Question:
|
Would you like to continue the telephone support in this form?
|
Answer:
|
Yes – no – don’t know – no answer
|
Question:
|
In the future, could this telephone support make contacts to doctors or psychologists less necessary or perhaps partly replace them?
|
Answer:
|
Yes – no – don’t know – no answer
|
Question:
|
Is there anything you would change or improve?
|
Answer:
|
Yes – no – don’t know – no answer and additional free text
|
Statistical analysis
The baseline characteristics were compared by group affiliation to identify any group differences at the baseline. We used t-tests for continuous variables and Chi-square tests for categorical variables. To analyse effects of the telemedical intervention on quality of life, the WHOQOL total score and each of the WHOQOL domains of the intervention group were compared with the scores of the control group after 6 months using t-tests.
The analyses were conducted with the intention-to-treat approach. For randomised clinical trials with missing data, the multiple imputation procedure is a valid method for handling missing data [19] and minimising possible biases [20]. However, a required condition for multiple imputation is that missing data is distributed completely at random (MCAR) or at random (MAR), whereas the method is less appropriate for data missing not at random (MNAR) [21]. After thorough inspection, we appraised the missing data as MAR. The proportion of missing values ranged from 11 – 17% (WHOQOL variables 12%). Therefore, multiple imputation was used. To be able to reproduce the results, the random seed value was specified each time the analysis was performed [19]. Eighteen variables were included in the imputation model. Minimum and maximum values for score values were defined. Further details are documented in the supplementary information file. All statistical procedures were performed in SAS 9.4 (© 2002-2012 by SAS Institute Inc., Cary, North Carolina, USA.).