Descriptive and Inferential Statistical Analyses

All the studies included in this review used both descriptive and inferential statistics for statistical analyses. Descriptive statistics are concerned with the presentation, organisation, and summarisation of data (Norman & Streiner, 1994). Common descriptive statistics used in patient satisfaction with treatments in CLBP studies included measures of central tendency such as mean, median, and mode, and measures of dispersion such as range, minimum, maximum, and standard deviation.

Inferential statistics enable generalisation from a sample of data to a larger group of individuals (Norman & Streiner, 1994). Common inferential statistics used in patient satisfaction with treatments in CLBP studies included regression analyses (See for example,Mannion et al., 1999), analysis of variance (See for example, Licciardone et al., 2003; Mannion et al., 1999; Nyiendo et al., 2001), and chi2 test (See for example, Licciardone et al., 2003).

3.4.7 Definitions and Conceptual Frameworks of Patient Satisfaction with Treatments in CLBP

Of the 26 studies included in this systematic review, none operationally defined patient satisfaction or dissatisfaction with treatments in CLBP, or proposed a conceptual framework or model of what comprises patient satisfaction with treatments. Often, there appeared to be no distinction between satisfaction and dissatisfaction, and the way that satisfaction and dissatisfaction were measured generally implied equality between the concepts. Further, none of the studies stated how the concepts were used, although sometimes this could be inferred. For example, no study discussed whether satisfaction was considered an independent or dependent variable, and studies did not state whether the concept was perceived as uni-dimensional or multidimensional but this information could be extrapolated from closer observation of the measures used or statistical analyses.

3.4.8 Questionnaires Used to Measure Patient Satisfaction with Treatment in CLBP All 26 studies reported satisfaction data from patients‘ perspectives, all using a patient-reported assessment. There were three studies that used a telephone interview which may have influenced results because for example, even with pre-specified responses it required interpretation of responses for coding (Nyiendo et al., 2000; Nyiendo et al., 2001; Carey et al., 2000).

Based on the findings of this systematic review, there is no consensus or ‗gold standard‘ for measuring treatment satisfaction or dissatisfaction in CLBP. The Cherkin and McCornack Satisfaction Questionnaire was used in 8% of studies (n=2) (Nyiendo et al., 2000; Nyiendo et al., 2001). This instrument covers aspects relevant to patients‘ satisfaction with care and contains 11 items (see section 3.4.9). In addition, one study (Pincus et al., 2000) adapted a 27-item osteopathic and GP management satisfaction questionnaire originally developed by Linn & Greenfield (1982). One important consideration is that the adapted 27-item questionnaire was originally developed in patients with chronic illness and further details are not provided regarding for example, types of chronic illness and number of patients, if any, with CLBP which is the focus of this thesis. Whilst there are some similarities between patients with CLBP and those with other chronic indications (such as that they may both experience pain which may cause some form of disability), there may be key differences. For example, someone with chronic pain induced from having for example, cancer, may have a very different experience of pain, or may rationalise their pain and any limitations they experience in the context of survival, whereas an individual with CLBP may not face issues related to survival and therefore may perceive their pain and its impact on daily activities very differently.

In addition, five studies used study specific questionnaires (Carey et al., 2000; Goodwin & Goodwin, 2000; Mannion et al., 1999; Molinari & Gerlinger, 2001; Rainville et al., 1997). See Table 4 for details on the content of satisfaction items or factors. In general, these study-specific questionnaires measured satisfaction with various components of treatment such aspects of the therapeutic relationship or a particular rehabilitation programme. Also, the content was usually unidimensional measurement and so many aspects of treatment satisfaction related to the focus of this thesis were omitted such as satisfaction with the treatment process (e.g. involvement in treatment decisions), and often

Further, satisfaction assessments differed in the number of items included, and the rating scales. Notable examples include studies that measured patient satisfaction modelled on the Cherkin & MacCornack satisfaction questionnaire and a 5-point Likert-type scale (Nyiendo et al., 2000; Nyiendo et al., 2001), and 13 studies that used a single global satisfaction item of various kind (Barker et al., 2008; Buchner et al., 2006; Buchner et al., 2007; Buchner et al., 2007; Barker et al., 2008; Hazard et al., 1994; Holm, Friis, Storheim, & Brox, 2003; Katz et al., 2005; Licciardone et al., 2003; Macario et al., 2008; Shirado et al., 2005; Smeets et al., 2006; Torstensen et al., 1998). Closer examination of the global satisfaction items revealed the diversity between studies using satisfaction ratings, since they focused on different aspects of CLBP and its treatment. For example, one global satisfaction rating was specific to pain relief (Katz et al., 2005) while another measured back satisfaction (Holm et al., 2003).

In fact, this potential problem was not just limited to global satisfaction ratings and CLBP studies appear to assess satisfaction with many different aspects ranging from pain relief (Katz et al., 2005), back satisfaction (Holm et al., 2003), satisfaction with the rehabilitation programme (Goodwin & Goodwin, 2000), satisfaction with therapy (Buchner et al., 2006; Buchner et al., 2007; Buchner et al., 2007; Mannion et al., 1999), and satisfaction with care (Carey et al., 2000). This makes comparisons between studies particularly difficult.

In addition, the response scales also varied for global satisfaction assessments from 5-point Likert-type scales (See for example, Buchner et al., 2006; Buchner et al., 2007; Buchner et al., 2007), to scales that ranged from 0-10 or 0-100 (See for example, Groessl et al., 2008; Haas et al., 2005) where higher scores indicate a level of satisfaction. Some global satisfaction assessments had 4 categorical options such as ‗very satisfied‘,

‗satisfied‘, ‗not satisfied and not disappointed‘, and ‗unsatisfied‘ (See for example, Shirado et al., 2005).

Comparing satisfaction results across these 26 studies is complicated because the studies did not focus on satisfaction and because the design of the studies and measurement of satisfaction varied so greatly. However, some key points did emerge when the satisfaction results were compared, as shown in Table 6. Most studies reported a positive level of satisfaction. Nevertheless, it is not apparent from this review whether these positive results were a true reflection of patients‘ scores or if some have ceiling effect (perhaps a consequence of the way studies measured satisfaction). Also, in general, chiropractic care appeared to be favoured by patients over medical care (Nyiendo et al., 2001; Nyiendo et al., 2000).

It should be noted that whilst there are numerous global ratings of satisfaction items used in CLBP studies, these generally require patients to average their opinion of for example, ‗satisfaction with therapy‘ into one single question at each timepoint. Chapter 2 demonstrates the various aspects that may be related to patient satisfaction with treatment (e.g. information/knowledge provided, involvement in treatment decisions, therapeutic relationship, patient perceptions and experiences with treatment etc.) and therefore the appropriateness of global ratings could be questionable depending on the intended purpose. For example, can patients truly average all of that information into a single item, and if they can, is there consistency between patients in terms of what they consider in making their global evaluation? Whilst global ratings are brief and easily administered, the inconsistency incurred when patients make their assessments possibly based on different criteria, may ultimately impact on reproducibility of scores (Feinstein, 1987). In addition, global ratings of satisfaction are subject to ceiling effects and can disguise or hide aspects of dissatisfaction (Lebow, 1974; Locker & Dunt, 1978). Further, global ratings are

generally known to be less informative than for example, disease-specific or treatment- specific questionnaires which are multidimensional (Locker & Dunt, 1978; Hudak & Wright, 2000). Consequently, global rating assessments are sometimes used to aid interpretation. This notion is documented in the FDA PRO guidance which recommends the use of global ratings in general (not specific to treatment satisfaction) to help interpretation by the definition of responders (Food and Drug Association, 2009). Therefore, although a global rating of treatment satisfaction may in general not be suitable as an only assessment, it may be helpful in some studies in addition to other measures to aid interpretation of results.

Table 6: Patient Satisfaction Data in Included CLBP Studies

Study* Positive (+) Satisfaction Data Neutral (0)

Satisfaction Data

Negative (-) Satisfaction Data

Barker 2008

In the FairMed group, 27% of patients stated that they were more able to cope with pain at 3weeks; in the TENS group, it was 45%.

73% of the FairMed participants stated no change in their ability to cope with pain compared to 44% in the TENS group.

11% of TENS group participants stated that they were less able to cope with pain at 3 weeks.

Buchner 2006

Using a 5-point Likert scale, the mean

satisfaction with therapy score for patients with CLBP was 2.85 (SD ± 1.61).

Buchner 2007

Using a 5-point Likert- type scale, satisfaction with therapy scores between the three age groups ranged from 3.10 to 3.48 (SD ± 1.43 to 1.58) at 6 month follow-up.Results between different age groups were not statistically significant. Buchner

2007

Using a 5-point Likert- type scale, satisfaction with therapy scores between the three groups of chronicity ranged from 3.13 to 3.45 (SD ± 1.40 to 1.58) at 6 month follow-up. Results between groups of chronicity were not statistically significant.

Study* Positive (+) Satisfaction Data Neutral (0) Satisfaction Data

Negative (-) Satisfaction Data

Carey 2000 At 22 months, patients were asked about their overall satisfaction with care. Care was rated as ‗very good‘ or ‗excellent‘ by 25% of patients with unremiting CLBP compared to 38% of those with remitting CLBP. Chown

2008

Results on the 5-point Likert-type scale were collapsed. At baseline, the majority of patients were

‗somewhat/very satisfied‘ with group exercise, physiotherapy and

osteopathy: 39%, 42%, and 40% , respectively. The proportion of patients stating ‗somewhat satisfied‘ or ‗very satisfied‘ with overall medical treatment increased for all treatment groups (group

exercise/physiotherapy/ osteopathy) between baseline and 6 weeks. 63%, 79%, 87%, respectively.

Goodwin 2000

The following elements of the programme scored positively - above 7 on 0-10 VAS scales: the

introduction, understanding back pain, pain theories, open discussion about pain, ergonomics, exercise principles, stress and relaxation, gym,

hydrotherapy, part of a group, physical abilities, psychological abilities, staff and organisation.

The following elements scored negatively - below 7 on 0-10 VAS scales: healthy back video, sleep and beds, and mentoring.

Groessl 2008

On VAS scales of 0-10, mean scores for the health benefits received from the yoga program, the yoga instructor, and the ease of participation were 5.97, 9.09, and 6.03, with higher scores indicating more satisfaction. Haas 2005 On a scale of 0-100, results indicated

that patients with CLBP receiving chiropractic care had significantly higher patient satisfaction than patients receiving medical care: mean 86.4 SD 19.9 vs. mean 71.3 SD 22.7, respectively, p<0.01.

Katz 2005 On a scale of 1 to, satisfaction with pain relief was 3.43 (SD 1.06) for patients receiving bupropion compared to 2.78 (SD 1.07) for patients receiving placebo. This difference reached statistical significance.

Licciardone 2003

Both osteopathic manipulative treatment (p=0.001) and sham manipulation (p=0.02) participants reported significantly greater satisfaction with their back care than the no intervention control

participants.

Study* Positive (+) Satisfaction Data Neutral (0) Satisfaction Data

Negative (-) Satisfaction Data

treatment (for spinal decompression) was 8.55 (median 9, range 5 to 10). Mannion

1999

The majority of the patients declared their satisfaction on hearing which group they had been assigned to, and few of them changed their impression for the worse during the course of the treatment. This was observed for all three groups.

Molinari 2001†

The non-operative group had an average satisfaction score of 9.6 of 15 (SD 2.8). The operative group had an average satisfaction score of 13.9 of 15 (SD 1.8). The difference in satisfaction scores between the groups reached statistical significance. Niemisto

2005

At 2 years, the combination group (receiving combined manipulation, stabilising exercises and physician consultation) had higher satisfaction with care compared to the

consultation group. Norris

2008

Mean values of all patient satisfaction questions showed positive experience (>4.5 points).

Nyiendo 2001

There was a sharp contrast favouring chiropractic in the proportion of patients that reported satisfaction with care at 1 year; the trend was apparent on all 10 satisfaction questions (p<0.0001). Differences between chiropractic and medical care were found in patients‘ confidence that the treatment was working (36% vs. 74%) and in the proportion of patients who would see a physician of the same type in the future for a CLBP problem (61% vs. 83%). For both groups, patients were least satisfied with ‗sufficient information provided about the cause of their pain‘ (40% vs. 73%).

Nyiendo 2000

Satisfaction was higher for patients attending chiropractors than medical physicians. In particular, patients expressed greater satisfaction regarding information on treatment program provided, and overall medical care.

Pincus 2000

Levels of satisfaction were high (for competence, quality of care, and efficacy) for GP management and osteopath; however, there were significantly higher scores for

satisfaction with osteopathic treatment compared to GP treatment in the same surgery.

Study* Positive (+) Satisfaction Data Neutral (0) Satisfaction Data

Negative (-) Satisfaction Data

involvement (e.g. patients receiving Workers‘ Compensation, Social Securuity Disability, or private disability policy benefits). Where items scores ranged from 1to 10 (excellent to poor), mean item and total satisfaction scores were similar between those with and without compensation involvement (16.4 and 16.7, respectively).

Shirado 2005

Eighty-five patients (48.6%) were satisfied with the back school 12 months after enrollment. Fifty-eight patients (33.1%) were satisfied.

Twenty patients (11.4%) were not satisfied/not disappointed (equal) with the back school 12 months after

enrolment.

Twelve patients (6.9%) were

unsatisfied/disappointed with the back school 12 months after enrolment.

Smeets 2006

Satisfaction scores for three different percentiles of the baseline Roland and Morris Disability Questionnaire (RMDQ) were presented. Satisfaction was significantly higher in the active physical therapy group compared to the waiting list control group when the patient had a lower level of functional limitations at pre-

treatment. For the ninetieth percentile score (RMDQ = 19) this difference was not significant. CBT and combined therapy showed a significantly higher level of

satisfaction compared to the waiting list group, and the higher the baseline RMDQ-score, the greater this difference became. No differences were evident between CT and CBT. Torstensen

1998

A total of 34.2% (26 patients) in the medical exercise therapy group (MET), 32.2% (19 patients) in the conventional physiotherapy (CP) group, and 6 patients 9.5% (6 patients) in the ordinary activity level group were ‗completey satisfied‘ with their treatment. Many patients were ‗satisfied‘ with their treatment: 28 in the MET group, 21 in the CP group, and 24 in the ordinary activity group.

There were 9 patients in the MET group, 14 in the CP group, and 25 in the ordinary activity group were ‗partly satisfied‘ with their treatment.

There were 4 patients in the MET group, 5 in the CP group, and 8 in the ordinary activity group who were ‗dissatisfied‘ with their treatment.

Wallace 2009

Atotal of 69% of the sample was completely satisfied with all elements of their care and 63% did not intend to seek care from another health-care provider.

Number of studies

21/24 5/24 5/24

*

Only first author reported

N.B. Hazard 2001/Holm 2003 are not included since papers document correlations/associations only.