Study design and setting
The study was a multi-center prospective observational cohort study. We recruited participants from one secondary and two tertiary referral centers between August 2017 and November 2018. All participants had been referred for operative treatment for symptomatic basal thumb joint OA that had not responded to non-operative treatment. Helsinki University Hospital Institutional Review Board approved the study protocol before commencement of the study. All methods were carried out in accordance with relevant guidelines and regulations.
Study hypotheses
In people being treated for basal thumb joint OA, our null-hypotheses were:
-
1)
The longitudinal validity of the PRWHE pain subscale and a single item pain NRS are not significantly different (measured as the correlation between target instrument and participant-reported global improvement or hand function improvement)
-
2)
The signal to noise ratio (expressed as a standardized response mean, SRM) of the PRWHE pain subscale is not significantly different from the NRS.
-
3)
The length of the recall period does not affect the longitudinal validity of the pain NRS (correlation between improvement in pain and participant-reported global improvement or hand function improvement)
-
4)
Difference in the length of the recall period does not affect the reported level of pain.
Participants
All people referred for basal thumb joint OA were screened by a board-certified hand surgeon. If they were dissatisfied with the pain and/or function of the hand, and accepted the risks related to surgery, standard surgical care (trapeziectomy) was offered and they were informed about the study. Recruitment occurred after the decision to operate was made.
Inclusion criteria
Exclusion criteria
-
Systemic inflammatory disease
-
Neurological disease that weakens upper arm muscles
-
Any operation of the same hand during the past 6 months
-
Bilateral operation planned
-
Zig zag deformity (> 45 degrees of thumb metacarpophalangeal (MP) joint hyperextension in rest)
Interventions
After recruitment, all participants were given an Actimove® Rhizo Forte splint (Essity AB, Stockholm, Sweden), and they were instructed to use it continuously for six weeks. After splinting period, all participants underwent a simple trapeziectomy. The trapezoid bone was exposed using a volar approach and removed in a piecemeal fashion. The space between first metacarpal and scaphoid was carefully examined for any bone remnants before the remaining joint capsule was closed. We did not use any interposition or suspensionplasty. After skin closure, the thumb was immobilized with a cast in palmar abduction/opposition for 3 weeks. At 3 weeks, the cast was removed and participants were provided with a post-operative exercise protocol that they could perform at home daily. They were also instructed to use the orthosis in daily activities as needed for the next three weeks and after that only if needed to control the pain.
Outcomes
The baseline characteristics were recorded after the participants had signed the informed consent. Questionnaires were sent by mail to all participants. The outcome measures were collected at baseline, after 6 weeks of splinting, and then 3 months, 6 months, 9 months and 12 months post-surgery.
Pain was measured by two instruments:
-
1)
a single item 11-point NRS where 0 = no pain and 10 = worst possible pain. The same question was posed three times with different recall periods: “How would you describe the pain in your thumb during the past day / during the past week / during the past month while doing daily activities, work and/or hobbies?”. The intraclass correlation coefficient (ICC) in NRS with knee OA patients is 0.95, whereas standard error of measurement (SEM) is 0.48 and minimum detectable difference (MDD) 1.33 [13]. The minimal important difference (MID) is 2.0 with chronic musculoskeletal patients [14].
-
2)
the PRWHE pain subscale which comprises five items as described above. The four items assessing pain intensity are measured on an 11-point NRS scale where 0 = no pain and 10 = worst possible pain. The recall period is one week. The fifth item assessing pain frequency is measured on an 11-point NRS scale where 0 = never to 10 = always. The pain subscale is the sum of the five items and ranges from 0 (best) to 50 (worst). PRWHE is the same instrument as the PRWE [15] but the word ‘wrist’ is replaced by ‘hand’. PRWE has been validated in Finnish [16].
All pain items were in the same order in the questionnaires: day, week and month.
The anchor measures in the assessment of longitudinal validity were
-
1)
participant-reported global improvement measured by a 5-point Likert scale (much better; little better; no change; little worse; much worse).
-
2)
the PRWHE function subscale which comprises 10 items rating the difficulty in doing specific tasks during the past week on a NRS scale from 0 = not difficult to 10 = unable to do. The PRWHE function subscale is the sum of the 10 items and ranges from 0 (no disability) to 100 (worst disability). Systematic review found an ICC of 0.85 for PRWHE [17], SEM was 2 points with various hand surgery patients, [18] MDD was 11 points, and minimal clinically important difference (MCID) was 11.5 points in patients with distal radius fracture [19].
Statistical analyses
Change scores were calculated by subtracting baseline score from the follow-up scores. We changed the direction of the pain scales so that a positive value corresponds with improvement in pain. We used measurements from all time points for the longitudinal validity assessment (50 target measure–anchor measure data pairs at five follow-up points = 250 data pairs), and also for the comparison of different recall periods of the single item NRS (6 time points * 3 question * 50 participants = 900 observations).
To compare the longitudinal validity of the three pain NRS recall periods (past day, week, month) and the PRWHE pain subscale, we calculated Spearman correlations between the target measures and participant-reported global improvement; and Pearson correlations with the target measures and change in PRWHE function subscale. Confidence intervals for the correlations were derived by bootstrapping 1000 samples. We then compared these correlations between measures using the method described by Steiger [20].
To assess the signal/noise to ratio, SRMs were calculated as change in score / SD of change for all pain outcomes using data from baseline and at 12 months after surgery. We compared the SRMs of the single item pain NRS scores and the PRWHE pain subscale scores using a “modified Jack-knife test” similar to Angst et al. [21].
We estimated the difference in the reported pain in different recall periods using a linear mixed model. Participant was entered as random factor and the question type (pain during past day, week, or month) and time points as fixed factors. We also fitted another model assessing if the time from recruitment modified the effect (time point*question interaction). Estimated marginal means from the mixed model were used as mean values. In all analyses, p-values less than 0.05 were considered statistically significant.