Captación y almacenamiento de aguas lluvias

Sub-group analyses were performed to examine possible causes of heterogeneity. The positive association between compensation and

unsatisfactory outcome was seen in all sub-groups, with little variation in the magnitude of the association. The results for each sub-group are summarised in Table 2.4, and forest plots are provided where indicated. The significance of the difference between subgroups was analysed by meta-regression and these results are provided below.

Analysis according to study type (Figures 2.4 – 2.6) showed a stronger association in randomised controlled trials, compared to either cohort studies or case series. However, there were only two randomised controlled trials in the analysis, which meant that the confidence interval for the effect estimate was wide and the difference seen between study types was likely to be due to chance. This variable was also analysed on meta-regression and the

difference was found to be not significant. Although the use of randomisation may indicate better methodology, it must be remembered that the exposure variable being analysed (compensation status) was not the variable

Table 2.4. Sub-group analyses of the association between compensation status and unsatisfactory outcome.

Subgroup Studies Odds ratio 95% C.I.

Study type

Randomised controlled trial 2 5.03 3.22 – 7.86

Cohort study 30 3.58 2.74 – 4.67

Case series 97* 3.87 3.25 – 4.61

Minimum time to follow-up

0 – 6 months 21 3.81 2.72 – 5.34 7 – 12 months 30 4.02 3.08 – 5.25 13 – 24 months 34* 4.36 3.17 – 6.01 Over 24 months 30 3.44 2.60 – 4.55 Completeness of follow-up 80% or more 111* 3.84 3.30 – 4.47 Less than 80% 18 3.61 2.39 – 5.47

Prospective versus retrospective

Prospective 15 3.60 2.70 – 4.80

Retrospective 114* 3.84 3.27 – 4.50

Procedure

Lumbar spine discectomy 24 4.77 3.51 – 6.50

Lumbar spine fusion 19 4.33 2.81 – 6.62

Shoulder acromioplasty 13 4.48 2.71 – 7.40

Carpal tunnel decompression 10 4.24 2.43 – 7.40

Lumbar intradiscal chymopapain injection 9 3.67 2.45 – 5.51

Country of origin

U.S.A. 106* 3.77 3.20 – 4.43

Canada 12 4.02 2.65 – 6.09

All Europe 6 7.42 4.37 – 12.60

Australia 5 2.23 1.49 – 3.35

Study designed to assess compensation effect

Yes 16 3.60 2.50 – 5.20

No 113* 3.85 3.29 – 4.51

Compensation type

Worker’s compensation only 86* 3.89 3.26 – 4.64

Worker’s compensation and litigation 43 3.69 2.88 – 4.73

Revision versus primary surgery

Primary surgery only 81 3.66 3.07 – 4.36

Revision surgery only 19 5.54 3.47 – 8.83

Figure 2.4. Forest plot of results according to the study type for randomised controlled trials.

Analysis according to minimum time to follow-up revealed similar odds ratios for each of the four groups, with overlapping of all confidence intervals. This was also found to be not significant on meta-regression. Analysis according to completeness of follow-up (less than 80%, compared to 80% or more)

revealed similar odds ratios and was also found to be not significant on meta- regression. Comparing prospective studies to retrospective studies also revealed similar odds ratios and was not significant on meta-regression.

Due to the association between compensation and injury, most studies dealt with orthopaedic, plastic and spinal surgery. The forest plots for the most common procedures are shown in Figures 2.7, 2.8, 2.9, 2.10, 2.11 and 2.12. The odds ratios for the six most common procedures (in order: lumbar discectomy, lumbar spine fusion, shoulder acromioplasty, carpal tunnel release, lumbar intradiscal injection of chymopapain, and cervical spine fusion) were similar. The differences in the effect for each of these treatment groups was not significant on meta-regression.

Analysis according to geographic origin showed a stronger association in European studies, and a weaker association in studies from Australia. Both of these groups, however, contained a small number of studies (Table 2.4, and Figures 2.13, 2.14, 2.15 and 2.16) and the differences were not significant on meta-regression.

Figure 2.7. Forest plot for subgroup analysis of lumbar discectomy.

Figure 2.8. Forest plot for subgroup analysis of acromioplasty.

Figure 2.10. Forest plot for subgroup analysis of intradiscal chymopapain.

Figure 2.11. Forest plot for subgroup analysis of carpal tunnel decompression.

Figure 2.14. Forest plot for subgroup analysis of studies from Europe.

Figure 2.15. Forest plot for subgroup analysis of studies from Australia.

The OR for studies looking only at worker’s compensation patients (not litigation) was similar to studies looking at patients treated under worker’s compensation or litigation (Table 2.4). There were no studies that examined only litigating patients (i.e., excluding workers’ compensation). There was no significant difference on meta-regression.

The OR for studies looking specifically at the effect of compensation (i.e., studies designed as a compensation versus non-compensation cohort) was similar to the OR in studies not specifically designed to examine the effect of compensation (Table 2.4).

Univariate meta-regression of the sub-groups analysed above found them not to be significantly associated with the effect of compensation on the risk of an unsatisfactory outcome. On meta-regression, the year of publication was found not to be significantly associated with the effect size. The effect of sample size, however, was such that the effect size decreased with increasing sample size of the studies.

A post hoc analysis was performed to assess the effect of revision versus primary surgery (Table 2.4). This showed the association between

compensation and poor outcome to be stronger in studies of revision surgery. Analysis of this variable by meta-regression, which included studies with mixed populations according to the ratio of revision to primary cases, showed this association to be strongly significant.

In general, the level of heterogeneity decreased in the sub-groups. Greater homogeneity was seen in the prospective studies (p = 0.03), randomised controlled trials (p = 0.46), and in the European and Australian studies (p = 0.68 and p = 0.67, respectively). Grouping by procedure also reduced

heterogeneity, however the heterogeneity remained high for other sub-groups. A funnel plot of all included studies (Figure 2.17) revealed some asymmetry in studies with larger standard errors (smaller populations).

Of thirteen studies reporting continuous outcomes, ten noted a statistically significant association between compensation status and poor outcome, and three noted no significant difference. This is consistent with the findings of the studies with dichotomous outcomes. As only four of these studies provided the means and standard deviations necessary to calculate standard mean difference, a meta-analysis was not undertaken for this group.

Figure 2.17. Funnel plot of all studies included in the meta-analysis.*

*Arrow denotes one point outside the scale. Vertical dashed line denotes summary odds ratio.