Impact of a nurse led telephone intervention on satisfaction and health outcomes of children with inflammatory rheumatic diseases and their families: a crossover randomized clinical trial

A randomized crossover, experimental longitudinal design was used in this study (see protocol published elsewhere) [20]. This article presents the quantitative component of this study.

The setting was a tertiary referral pediatric rheumatology outpatient clinic, serving all French-speaking cantons of Switzerland. Every year, about 110 new patients are admitted to the clinic; about 50 of them have chronic inflammatory rheumatic diseases.

The study participants were the designated users of the nursing telephone consultation, therefore included parents of children ≤11 years or children from 12 years of age. Children newly diagnosed (within 18 months prior to the enrolment date) with an inflammatory rheumatic disease, including JIA, connective tissue disease, and vasculitis and under the age of 16 at enrolment or their parents were eligible. Upon agreement to study participation, informed consent and witten assent were provided by parents and children (aged between 11 and 16), respectively. Potential participants that did not understand and speak French and/or had no access to a telephone were excluded.

The study and its amendment were approved by the Human Research Ethics Committee of the canton of Vaud, Switzerland on January 17, 2011 and March 28, 2011, respectively. Parents and patients who attended the pediatric rheumatology outpatient clinic between January 2010 and August 2012 and consented were enrolled in the study for a total of 24 months. Briefly, this study was a randomized, crossover trial, in which patients were their own control [21]. The intervention (TN) was evaluated against standard care (SC) with the same subjects. It is worth noting that TN was provided in addition to SC, thus all participants received SC for the whole duration of this study. Participants were randomized and allocated to group 1 or group 2 using a computer-generated simple block randomization to account for different level of severity of illness. Treatment allocation was in sealed numbered envelopes. Both groups received 12 months of TN and 12 months of SC, only the attribution order varied; (group 1 received TN first and then SC and group 2 received SC first and then TN).

The Cox’s Interaction Model of Client Health Behavior (IMCHB), which was developed to direct and document nursing evaluation and care and reach positive nursing intervention effects on health outcome, was used to guide this study [22]. The main objective of this nurse-led intervention was to ensure continuity of care for children and their families. TN provided by specialized nurses via telephone included provision of individualized health information, affective support and help in decision making. Two qualified specialist nurses with over five years of experience in adult and pediatric rheumatology were specifically trained (3 day course) in TN oral communication, strategies for questioning parents and adolescents, assessing the quality of interactions and aiding decision making for the TN. A two-part standardized form of telephone interviewing was developed for each TN consultation. The first part included description of the call, such as time, initiator and nature of the call, action/decision taken, and a brief summary of the conversation and planned action. The second part related to the intervention itself and included eight questions on: 1) everyday life, school and social, 2) treatment, 3) physiotherapy, 4) occupational therapy, 5) pain, 6) schedule, 7) administrative issues, 8) any additional topic that the respondent would like to discuss.

As per cross-over design, all participants received the intervention (TN), either during the first 12 months or the last 12 months of the study. When in the TN group, participants attended a first face-to-face medical and nursing consultation at the start of TN (T0 for group 1; T12 for group 2). This visit allowed the TN nurse to introduce herself, explain how the telephone consultation would be carried out and get to know the child’s clinical, social and family situation. For the following 12 months, the participants received a monthly telephone call. In addition, the participating parent or child was given a telephone number to contact, when needed, the TN nurse on duty during normal office hours on week days.

As per cross-over design, all participants included were part of the control group, either during the first 12 months or the last 12 months of the study. Participants in the control group received SC, in which medical management was provided by a pediatric rheumatologist mainly, but also by other specialists (occupational therapists) as determined by patients’ needs. When in the SC group, participants attended a face-to-face medical consultation only, at the start of SC (T12 for group 1, T0 for group 2). The medical consultation was repeated every three months and participants were followed and treated as per standard practice. Parents were also informed that they could call the outpatient clinic.

The choice of data collection points and the study time span was based on theoretical and practical considerations [23]. Face-to-face consultation baseline data (demographics, health status, and satisfaction) were collected at T0. Demographic data about participants included age, gender, cultural background, marital status, occupation, education, language spoken at home, and types of treatment. Collection points occurred every three months for disease activity and health status assessment (T3, T6, T9, T12, T15, T18, T21 and T24) and every six months for satisfaction (T6, T12, T18 and T24).

The study’s primary outcome was participants’ satisfaction (child/parent). Satisfaction was assessed using the Client Satisfaction Questionnaire-8 (CSQ-8), an 8-item version of the 18-item CSQ [24]. It is brief to administer, has good psychometric properties, and has been translated and validated in French. Each item of the CSQ-8 items is rated on a 4-point Likert-type scale giving a total score ranging between 8 (no satisfaction) and 32 (total satisfaction), a score ≥ 30 indicates satisfactory rating of satisfaction and a score < 30 a lack of satisfaction. Test comparison was the proportion of subjects who were satisfied in each group as well as changes in satisfaction scores within groups (between T0 and T12).

Secondary outcomes were clinical health status measurements performed every three months as per standard practice, using the Juvenile Arthritis Multidimensional Assessment Report (JAMAR) French version, of which original version was translated and validated [25‐27]. The JAMAR includes 15 self-reported measures that assess well-being, pain, functional status, health-related quality of life, morning stiffness, disease activity, disease status and course, joint disease, extraarticular symptoms, side effects of medications, therapeutic compliance, and satisfaction with illness outcome. For children too young to self-report, the parent version of the JAMAR was used. In this study, the items of interest were: (1) Assessment of functional ability through a 15-item questionnaire, in which the ability of the child to carry out daily living activities is scored: 0 = without difficulty, 1 = with difficulty, 2 = unable to do. A total score of 0 was considered as no difficulty in functional ability and a total score of ≥1 was considered as having some difficulty; (2) Rating of the intensity of child’s pain on a 21-point visual analogue scale (VAS) (0 = no pain; 10 = extreme pain). A total score of ≤3 was considered as no pain and a total of ˃3 was considered as having pain; (4) Assessment of morning stiffness was a “yes-no” item; (5) Assessment of extraarticular symptoms was two “yes- no” questions assessing fever and rash; (7) Rating of disease status at the time of the visit as remission, continued activity, or relapse; (8) Rating of disease course from previous visit as much improved, slightly improved, stable/unchanged, slightly worsened or much worsened (improvement of disease status was assessed when participants answered “remission” at item 7 and “much improved” or “slightly improved” at item 8); (14) Assessment of health-related quality of life was performed through a 10-item questionnaire having two dimensions, physical health and psychosocial health, composed of 5 items each. The responses were “never” (score = 0), “sometimes” (score = 1), “most of the time” (score = 2), and “all the time” (score = 3). Separate scores for the physical and psychosocial subscales can also be calculated. A total score of 0 was considered as no difficulty in quality of life and a total of ≥1 was considered as having difficulty in quality of life. Same quotation was applied to the subscales.

The study’s secondary comparisons were the proportion of subjects that had: i) no morning stiffness; ii) no pain; iii) no difficulty in functional capacity; iv) in remission; v) no difficulty in physical quality of life and vi) no difficulty in psychosocial quality of life in both groups.

Sample size and power were predicted based on the number of newly diagnosed children with inflammatory rheumatic diseases admitted to the study hospital’s paediatric rheumatology outpatients 2008 (N = 48). We anticipated that around 70 children would be admitted to the clinic in the 18-month screening period prior to enrolment into the study and considered that 80% of patients/parents would give consent to their participation (N = 56).

A power analysis was calculated based on the number of participants expected to complete the study, not the number recruited initially. For 50% difference in the proportion of subjects with a satisfaction score ≥ 30 (cut-off score) between the two groups, 23 subjects per group was required to reach a power level of .90 for an alpha level of .05 (two-sided test). To compensate for an expected attrition rate of 20%, we aimed to recruit 28 subjects in each group (total of 56 subjects).

An intention to treat analysis was performed. Random intercept mixed effect linear models were used for continuous outcomes and random intercept logistic mixed models for binary outcomes. The models tested the effect of treatment (TN or SC), period (year) and the interaction between treatment and period. Prior to data analyses, data were screened for data file’s accuracy, missing data, outliers, and distribution [28]. Data analyses were performed using Stata version 13 software (StataCorp LP, College Station, TX, USA).

Figure 1 summarizes the recruitment and group’s allocation. Of 711 children initially screened, 120 were eligible and 55 (46%) consented to participate and were included. Participants were randomized and allocated to group 1 that received TN the first 12 months and then SC (n = 30) or group 2 that received SC first and then TN (n = 25). After 12 months, one participant of group 2 withdrew from the study. After 24 months two additional participants withdrew from group 1.

Table 1 shows demographics of respondents at baseline. Responding children (n = 24, females 58.3%) had a mean age of 13.1 years, most of them were born in Switzerland and were still in school (87.5%). Responding parents (n = 31, 96.8% females) were mostly Swiss (71%), working (90.4%) and married (74.2%). Table 2 shows children clinical characteristics with the majority of them diagnosed with some form of JIA (70.5%), predominantly with juvenile enthesitis-related arthritis (ERA) (29%) or with oligoarticular JIA (27%). Other diagnosis included; uveitis (5.4%), chronic osteomyelitis (3.6%), chronic infantile neurological cutaneous articular (CINCA) syndrome (1.8%), lupus (1.8%), Crohn’s disease (1.8%), Behçet syndrome (1.8%), auto-inflammatory disease (1.8%) and juvenile dermatomyositis (JDM) (1.8%), with a few undetermined conditions (7.3%). Disease severity was assessed, by the treating physician using the Juvenile Arthritis Disease Activity Score (JADAS) as per standard practice and showed that most children had mild to moderate condition (90%) versus severe (10%).

Proportions of participant who were satisfied (CSQ-8 score ≥ 30) are reported in Table 3. At T12 and T24, the interaction between the treatment and the year was not significant. A model without this interaction compared the TN and SC impact on satisfaction independently of the year it was received and showed that probability of being satisfied (satisfaction scores ≥30) was 8 times higher at the End of the TN period when compared to SC (OR = 7.7, 95% CI: 1.8–33.6). Satisfaction scores progressively increased by 20% from T0 to T12 in the TN group. An opposite negative trend was observed in the SC group, where satisfaction progressively decreased by 60% between T0 and T12.