Sævar B. Gudbergsson, MPH, PhD1, Alv A. Dahl, MD, PhD2,4, Jon Håvard Loge, MD, PhD3,5, Lene Thorsen, PhD2, Line Oldervoll, PhD6,7 and Ellen K. Grov, RN, PhD8,9

From the 1Montebello-Center, The Norwegian Resource Center for Coping with Cancer, Mesnali, 2National Advisory Unit on Late Effects after Cancer Treatment, Oslo University Hospital, Radiumhospitalet, 3Chair, Regional Centre for Excellence in Palliative Care, Oslo University Hospital, 4The Faculty of Medicine, 5Department of Behavioural Sciences in Medicine, University of Oslo, Oslo, 6Department of Social Work and Health Science, Norwegian University of Science and Technology, Trondheim, 7Research and Development Group, LHL Clinics, 8Oslo and Akershus University College of Applied Sciences, Oslo, and 9Sogn og Fjordane University College, Førde, Norway

OBJECTIVE: This focused review examines randomized controlled studies included by the term “cancer rehabilitation” in PubMed. The research questions concern the type of interventions performed and their methodological quality.

DESIGN: Using the Medical Subject Headings (MeSH) terms: neoplasm AND rehabilitation, all articles with randomized controlled studies that included adult cancer patients, written in English, were extracted from PubMed. Papers covering physical exercise, psychiatric/psychological treatment or social support only were excluded as they had been reviewed recently. Abstracts and papers were assessed by 3 pairs of reviewers, and descriptive information was extracted systematically. Methodological quality was rated on a 10-item index scale, and the cut-off for acceptable quality was set at ≥ 8.

RESULTS: A total of 132 (19%) of the 683 identified papers met the eligibility criteria and were assessed in detail. The papers were grouped into 5 thematic categories: 44 physical; 15 art and expressive; 47 psycho-educative; 21 emotionally supportive; and 5 others. Good quality of design was observed in 32 studies, 18 of them uni-dimensional and 14 multi-dimensional.

CONCLUSION: Published randomized controlled studies on cancer rehabilitation are heterogeneous in terms of content and samples, and are mostly characterized by suboptimal design quality. Future studies should be more specific and well-designed with sufficient statistical strength.

Key words: rehabilitation; cancer patients; randomized controlled trials; PubMed, review.

J Rehabil Med 2014; 46: 00–00

Correspondence address: Sævar B. Gudbergsson, Montebellocenter, Kurstedvegen 5, NO-2610 Mesnali, Norway. E-mail: savar.gudbergsson@montebello-senteret.no, berg-g@online.no

Accepted Aug 14, 2014; Epub ahead of print Oct 7, 2014

INTRODUCTION

With ongoing improvements in prognosis for major cancer types from the early 1970s, the rehabilitation of cancer patients (RCPs) has become of clinical importance. An accepted definition of RCPs is provided by DeLisa in 2001: “Cancer rehabilitation is a concept that is defined by the patient and involves helping a person with cancer to obtain maximum physical, social, psychological, and vocational functioning within the limit by the disease and its treatment” (1, p. 970). The World Health Organization’s (WHO) report on disability subsequently presented a wider definition of rehabilitation as “a set of measures that assist individuals who experience, or are likely to experience, disability to achieve and maintain optimal functioning in interaction with their environments” (2, p. 96).

Typically RCPs occurs for a specific period of time, and can involve both single and multiple interventions delivered by a single professional, or a team of rehabilitation workers. RCPs may be needed from the acute or initial phase immediately after recognition of cancer as well as later on in the post-acute and maintenance phases.

With these definitions, this study reviewed the content, results, and methodological quality of the randomized controlled trials (RCTs) on RCPs in PubMed. A further reason for carrying out this review was the recent request by Alfano et al. (3) concerning revitalization of the link between cancer survivorship and cancer rehabilitation, and their presentation of a new model of comprehensive cancer rehabilitation involving a multidisciplinary team of providers (3). At the same time a closely related request was raised from both Nordic and European quarters (4). Finally, a recent Cochrane review (5) of multidimensional rehabilitation programmes for adult cancer survivors had reached the same conclusions as Alfano et al.

The first RCT of RCPs was noted in PubMed in 1979, and at the end of 1989, 21 RCTs had been recorded in that database. However, the annual number of studies has increased and, as of June 2012, a total of 616 RCTs concerning RCPs are cited in PubMed. The PubMed database is delivered by the US National Library of Medicine, it is free of charge and easily available, and therefore widely used as a tool by clinicians and clinical researchers. Since RCTs have the highest evidence level and the number of such papers found was more than 600, we decided to study only RCTs of RCPs published in English and registered in PubMed from 1990 to 2011. Our study should be considered as a focused review not fulfilling all the specific methodological demands needed for a systematic review according the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) or Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) designs. We reviewed papers on RCPs using RCTs with the aim of answering the following research questions: (i) What kinds of interventions have been published? (ii) What characterizes the quality of the research designs used in these studies?

METHODS

Literature search

A search of PubMed was carried out in order to identify relevant articles related to RCPs. “Cancer rehabilitation” is not a Medical Subject Headings (MeSH) term, and therefore neoplasm was used as the main MeSH term, covering all cancer diagnoses, and rehabilitation as the second MeSH term. By using only 2 MeSH terms in combination the search was broadened optimally. The following limitations were also defined: studies of humans, RCTs, all adults aged ≥ 19 years, from 1 January 1990 to 31 December 2011, published in English.

Selection criteria

Studies with interventions involving medical and surgical treatments and procedures were excluded. Due to recent reviews, we also excluded interventions concerning the following themes: physical exercise alone, psychiatric/psychological treatment, social support only, or artificial nutrition only (6–9). Observational and case-control studies without randomization were excluded due to the number of RCT papers.

Reviewers’ evaluating procedures

Six reviewers operated as 3 pairs of evaluators, and all reviewers held PhDs in oncology, 2 as doctors, and 2 as instructors in physical activities, 1 as a nurse, and 1 as a social worker. First, all abstracts were distributed at random (every third paper to each pair) between 3 pairs, and they were scanned for fulfilment of inclusion and exclusion criteria. If the criteria were unclear in the abstract, the full paper was examined. Each pair of reviewers resolved any disagreement between themselves by discussion. If they still disagreed the study was evaluated by one of the other pairs. Each pair read the full-text papers allotted to them by included abstracts, and eventual disagreements in the evaluations were settled in the same way as for the abstracts.

A registration form for major variables was developed, and data were extracted from each paper and rated according to predefined categories. Collected data were stored on separate extraction sheets for each paper, and then merged.

Identification of papers

The PubMed search identified 683 RCT papers, and based on the selection criteria 414 (61%) of the retrieved abstracts were excluded from further review. A set of 269 (39%) full-text papers was then left for evaluation, and 137 (51%) of these papers were excluded since closer inspection showed that they did not meet our selection criteria. This examination left 132 full-text papers for intensive studies (19% of the initially identified papers).

RESULTS

Characteristics of the randomized controlled trial papers

The characteristics of the studies were based on the total numbers of participants and, in all subgroups, the mean age, gender, and diagnosis were specified.

Of the 132 papers, only 11 (8%) used the term “rehabilitation” in the title. The number of participants in the groups varied from 10 to 921, with a total number of 16,331 reporting 184 different types of interventions (Table I). There were 40 studies with 130 or more participants (at least 65 in each group, which is the minimally necessary sample size when 2 groups are compared) and 92 with fewer than 130 participants (number not shown in tables). Only 16 (12%) studies were published between 1990 and 1999, 30 (23%) between 2000 and 2004, and 86 (65%) between 2005 and 2011.

Study interventions

Based on their content and interventions the studies were grouped into 5 thematic groups (Table I), as follows:

I. Physical interventions included physiotherapy, endurance exercise and strength, flexitouch, massage, elastic sleeve, finger acupressure, lymph drain, pelvic muscle exercise, reflexology, shoulder exercises, vacuum erection devices, yoga, and relaxation. This category contained 44 studies (33%) and 56 interventions (30%),

II. Art and expressive interventions concerned music therapy, art therapy, dance and movement, expressive writing, imagery, and reading, with 15 studies (11%) and 17 interventions (9%).

III. Psycho-educative interventions included self-care instructions, cognitive behavioural treatment, sleep education, and stress management, relating to 47 studies (36%) and 70 interventions (38%).

IV. Emotionally supportive interventions covered support groups, emotional support, support for family, and hope intervention programme with 21 studies (16%) and 33 interventions (18%).

V. Other interventions (n = 5) included food and nutrition (fish-oil, diet), lifestyle interventions, hypnosis, and smoking cessation contained by 5 studies (4%) and 8 interventions (5%).

Assessment of scientific quality

The rating of the methodological quality of the papers was based on the criteria published by Iles et al. (10, 11). Their approach included scoring of 10 quality items for each paper (Appendix I). Based on the scorings of the 2 reviewers, each paper was rated within a range of scores from 0 (poorest quality) to 10 (best quality). Since we mainly were interested in studies with good design quality, we applied a cut-off at the 75th percentile with total score of ≥ 8 as definition of “good quality” (GQ) paper. Accordingly, papers with scores < 8 were defined as “less good quality” (LGQ) papers.

Based on this dichotomy, 100 papers (76%) were classed as LGQ and 32 (24%) studies as GQ. Only the latter papers are described in further detail here.

The 32 GQ papers belonged to the following thematic groups: 9 to group I (3 also used interventions from other categories), 1 to group II, 15 to group III (2 also used interventions from other categories), 6 to group IV, and 1 to group V.

Characteristics of good quality studies

The 32 GQ studies assessed 50 interventions, and the characteristics of these interventions were as follows: 21 studies concerned 1 intervention (uni-dimensional) compared with standard or usual care, 11 studies compared 2 or more types of interventions (multi-dimensional) and, among them, only 4 compared the interventions with standard care (Table II). Three studies got maximum GQ ratings: Korstjens et al. (16) and Sharp et al. (18) in group I, and the study by Kissane et al. (40) in group III (Table II).

The GQ group consisted of 18 uni-dimensional (intervention compared with treatment-as-usual) and 14 multi-dimensional (2 or more interventions compared). The study by Sharp et al. (18) was uni-dimensional, while those by Korstjens et al. (16) and Kissane et al. (40) were multi-dimensional.

Other characteristics

As shown in Table II 17 of the studies assessed a mixed group of cancer diagnoses and 12 studies assessed interventions for breast cancer, 2 for prostate cancer, 1 study covered colorectal cancer and another lymphomas, each with both genders.

Respondents of both sexes were found in 16 studies, 13 studies examined females only, and 3 studies reported on males only.

A similar number of studies took place at the time of primary treatment (n = 12) and the time after primary treatment (n = 15), while 4 studies were conducted during and after primary treatment, and for 1 study the time of assessment could not be identified (Table II). Only 1 of the papers was published in the 1990s (20), 8 between 2000 and 2004, and 23 between 2005 and 2011. Close to half of the GQ studies were from the USA (n = 14), 13 from Europe, and 5 from other parts of the world (Table II).

Measurements used in the good quality studies

Several instruments were used to assess the interventions used in the GQ studies, and we grouped these instruments into 2 main categories of measurements: (i) instruments used to assess well-being as outcome were used in 16 studies (quality of life, lifestyles, social relationships, psychological and cognitive measurements, etc.); (ii) instruments used to assess functional outcome were used in 15 studies (shoulder movement, physical activity, physiotherapy with lymphoma, etc.). Of these 32 studies, well-being instruments were used in 16 studies, functional instruments in 1 study, and a combination of both in 15 studies.

DISCUSSION

The main finding of this review is that papers registered in PubMed published as RCTs on cancer rehabilitation are hetero­geneous in terms of samples and outcomes. The studies are mostly characterized by “less than optimal” design quality. Due to the heterogeneity of the research design, numbers of participants, genders and measurements, it was not feasible to perform any systematic or meta-analysis, hence we report the findings in a focused manner. Therefore, we also recommend that the design of future studies of RCPs should be more specific, multidimensional and well-planned.

This review, exploring RCTs studies of RCPs registered in PubMed, included full-text reading of 132 RCTs. Thematically these papers were divided into 5 groups, among which the groups of Physical and Psycho-educational interventions were the largest, with 44 and 47 studies, respectively. Groups II–IV have content focusing on general lifestyle improvement, increased well-being or reduction of “distress”, defined as a unpleasant emotional experience of psychological (cognitive, behavioural, emotional), social, and/or spiritual nature (12), while group I mostly focuses on functional measures.

Only 32 of the studies reach GQ score concerning design. A common weakness in many studies was the lack of description or explanation of the baseline disability, which the study wanted to change by its interventions. The authors seem to presuppose that all cancer patients have similar problems, mostly with lifestyle issues.

Another common weakness was the absence of statistical power considerations concerning sample sizes, which is of crucial importance for interpretation of group comparisons concerning outcome variables. Our analyses showed that 92 (70%) of the studies had less than 130 participants usually needed for safe comparison of 2 groups (65 × 2). If studies with smaller group sizes show significant statistical group differences, they are of clinical significance, but there is a considerable risk of lacking significant differences due to small sample size (type II statistical error).

Based on the explanations given by the authors of the papers evaluated, most of them focused on supportive interventions for cancer patients rather than rehabilitation, and almost half of the studies were conducted during the period of primary cancer treatment. Most of the authors claimed to fulfil the criteria of rehabilitation interventions, although they did not fulfil the definition of RCPs given by DeLisa (2, p. 96). We find it challenging that lifestyle interventions are classified as RCPs, although we admit that the WHO definition accepts as rehabilitation all interventions that reduce disability to achieve and maintain optimal functioning in interaction with the environment.

Most of the authors stated that their interventions improved health for the participants during the study period. On the other hand, the long-term effects of the interventions are unknown, since the majority of studies had either no follow-up or just a short follow-up period. The samples frequently showed selection bias and since they regularly lack attrition analyses, their external validity is open for discussion.

Using an index score for evaluating the quality of the studies, we categorized only 24% as GQ studies. This means that 3 out of 4 studies had considerable problems, either of design, material, methods or statistics. We consider this to be a high proportion, but we do not have data from related fields of investigation. Our findings also support the recommendations reported by Scott et al. (5) in their systematic review of multidimensional rehabilitation programmes, namely that researchers designing RCTs for RCPs should be more aware of methodological issues in the future. In addition there is a need to report more systematically and in more detail on sampling, statistical power, attrition, as well as diseases and treatment characteristics, such as time from diagnosis to interventions, cancer treatment received, and disease and treatment status during the intervention period.

Furthermore, functional impairment, assessment tools, and inclusion and exclusion criteria were less than optimally described and could definitely be improved and become more standardized. These suggestions are in line with the recent papers on the Nordic and European perspective on RCPs (3) and from the USA (2).

We may speculate whether the heterogeneity of interventions classified as RCTs in PubMed may be consequences of the rather wide and unspecific definitions of RCTs presented in the Introduction. The inclusiveness of the WHO definition of rehabilitation supports such heterogeneity (11), even if that definition requires disability as the basis for rehabilitation. The content of this definition states that the aim of rehabilitative interventions is to mobilize the patient’s optimal functional level to participate in the community.

In many of the studies reviewed there is lack of description of the levels of impairment before the start of the interventions. The main idea of many interventions seems to be secondary prevention addressing risk factors for a future disease burden caused by the cancer and/or its treatment, rather than to deal with defined needs for rehabilitation due to limitations in physical, psychological, social functions, or their combination.

The WHO definition indicates that rehabilitation can address a variety of functions with a consequent need for several types of interventions, and our findings based on the PubMed database confirmed such a plurality. On the other hand, this plurality may be considered problematic, since RCPs thereby loses more of its cancer-specific content. In addition, many cancer patients have complex functional impairments, which require a combination of rehabilitative efforts, and we found very few studies addressing such combinations.

Another perspective is related to PubMed methods of classification and characterization of the content of RCPs. This might be one explanation for the inclusion of studies in our search that barely concern rehabilitative interventions at all.

Study imitations and strengths

Using PubMed as the only database is a limitation of our study. However, PubMed is a major literature base in medicine, and frequently used, since it is free and easily accessible, and it is therefore worth exploring. The review must be considered as focused rather than systematic, identifying RCTs that will be consulted by clinicians. Coverage of studies in English only may be considered a limitation, since relevant papers could have been published in other languages.

It was not possible to report the effectiveness of all the interventions, since different studies concerned different groups of cancer patients, sexes, age groups, and times in the cancer trajectory. For example, among studies offering the same type of interventions, some reported socio-demographics and detailed cancer information, while others did not. Due to limited methodological descriptions in the studies reviewed, we have hardly been able to discuss the long-term positive or negative impacts of the interventions on the cancer patients’ impairments.

Conclusion

This review highlights those RCTs under the heading of RCPs in PubMed and recognizes that they cover a heterogeneous set of uni- and multi-dimensional interventions that we classified into 5 thematic groups. We observe that these interventions are more focused on secondary prevention, lifestyle, and supportive care than on rehabilitation in the strict sense. Based on our design quality index, only 24% of the included papers reached “good quality” concerning research design and metho­dology. We therefore recommend that future studies of RCPs should assess more specific factors related to the rehabilitation of cancer patients.

References