Table 2.
Multilevel regression analysis of the number of physiotherapy session a
| Model coefficients c | ||
|---|---|---|
| Factors b | Estimates (i.e. semi-elasticities) | 95% confidence interval |
| Patient needs | ||
| Age and gender | ||
| [20-39] men | .088 | .062- .114 |
| [40–59] men | .217 | .193- .242 |
| [60–79] men | .260 | .235- .285 |
| [80 + ] men | .211 | .175- .247 |
| [20-39] women | .095 | .071- .120 |
| [40–59] women | .232 | .208- .255 |
| [60–79] women | .291 | .266- .315 |
| [80 + ] women | .259 | .230- .288 |
| Conditionsd | ||
| Shoulder operation | .324 | .257- .391 |
| Total knee prosthesis | .252 | .203- .301 |
| Muscular disease | .175 | .107- .242 |
| Neurological diseases | .081 | .053- .109 |
| Other orthopedic surgery | .043 | .011- .074 |
| Osteoporosis | .035 | .008- .062 |
| Rheumatic conditions | .024 | .002- .047 |
| Other surgery | -.043 | -.067- -.019 |
| Lymphatic problems | -.055 | -.106- -.004 |
| Rehabilitation | -.084 | -.157- -.011 |
| Comorbidity burden (annual costs 103 CHF) | ||
| 1,001–2,000 | .020 | .006- .033 |
| 2,001–5,000 | .065 | .053- .077 |
| 5,001–10,000 | .130 | .115- .144 |
| 10,001–20,000 | .180 | .162- .198 |
| 20,001–50,000 | .223 | .199- .246 |
| >50,000 | .381 | .340- .421 |
| Provider practices | ||
| Variation coefficient of the physician (by quintiles rank) | ||
| 2nd | .043 | .028- .058 |
| 3rd | .062 | .046- .078 |
| 4th | .062 | .046- .078 |
| 5th | .082 | .065- .099 |
| Variation coefficient of the physiotherapist (by quintiles rank) | ||
| 2nd | .019 | .000- .038 |
| 3rd | .037 | .018- .057 |
| 4th | .065 | .044- .086 |
| 5th | .087 | .067- .107 |
| Funding regulations | ||
| Deductibles (CHF) | ||
| 601–1,500 | -.060 | -.076- -.044 |
| >1,500 | -.086 | -.115- -.058 |
| Ceiling per prescription (proportion of treatments by physician as a nine-session episode by quintiles rank) | ||
| 2nd | .080 | .064-.097 |
| 3rd | .154 | .137- .171 |
| 4th | .205 | .188- .223 |
| 5th | .235 | .216- .254 |
| Ceiling per prescription (proportion of treatment by physiotherapist as a nine-session episode by quintiles rank | ||
| 2nd | .081 | .059- .103 |
| 3rd | .114 | .092- .136 |
| 4th | .155 | .132- .177 |
| 5th | .227 | .202- .252 |
| Proportion of new treatments by physician by quintiles rank | ||
| 2nd | -.050 | -.066- -.033 |
| 3rd | -.058 | -.074- -.041 |
| 4th | -.095 | -.112- -.078 |
| 5th | -.154 | -.171- -.137 |
| Proportion of new treatments by physiotherapist by quintiles rank | ||
| 2nd | −0.045 | -.064- -.025 |
| 3rd | −0.071 | -.091- -.051 |
| 4th | −0.081 | -.101- -.060 |
| 5th | −0.104 | -.125- -.082 |
| Context variables e | ||
| Being treated by more than one physician | -.083 | -.122- -.045 |
| Being treated by more than one physiotherapist | .060 | .015- .105 |
| Residency canton (13 dummy variables) | ||
| Only one had a significant coefficient | .042 | .006-.078 |
| Model constant | 1.732 | 1.688-1.776 |
aAnalysis was restricted to the patients’ first treatment.
bOnly variables with a coefficient significantly different from 0 (p < 0.05) are shown.
cThe model coefficients can be interpreted as semi-elasticities, i.e. the percentage change of the outcome after a unit change of the explanatory variable. The reference category was: men, < 20 y, no condition identified from drugs prescriptions, inpatient diagnoses or a specialist contact, deductibles < CHF 600, being treated by one physician and one physiotherapist in the lowest quintiles rank for all their variables, and not resident in a large city.
dThe following conditions had no significant effect: cancer, mental conditions, back problems, other knee problems, total hip prosthesis, other hip surgery, other trauma, and other medical condition.
eUrban residence had no significant effect.