Article Figures & Data
Figures
- Figure 1. Differential attrition. Note: The figure shows estimates from model 1–3 contrasting patient with and without a named GP 1–4 years after the index date. Patients with an index date in 2015 lack data on mortality for t+4; otherwise, all models include the full study population.
PCC = primary care centre.
Tables
Patient 1 Patient 2 Index date (DD/MM/YYYY) 01/02/2013 01/02/2013 Last date of follow-up 31/01/2017 31/01/2017 Censored No Yes Censoring date N/A 31/01/2015 Effective follow-up 4 years 2 years The table illustrates the outcomes computed for two fictional patients with the same index date, one of whom was censored (because they died) during the follow-up period. Outcomes (number of visits, continuity of care) are counted over the full follow-up period for both patients, but the effective follow up is shorter for patient 2 because she is only observed for 2 years after the index date.
Named GP (n = 23 762) No named GP (n = 42 301) A. Before balancing B. After balancing Variable Mean Variance Skewness Mean Variance Skewness Mean Variance Skewness Age, years 47.4 4422 26 44.2 5468 24.1 47.4 4424 26.1 Education 0.889 0.605 0.194 0.939 0.65 0.111 0.889 0.605 0.194 Disp. inc. (decile) 5.64 8.17 -0.0662 5.36 8.86 0.0283 5.64 8.17 -0.0604 Born in Sweden or Nordic region 0.883 0.103 -2.39 0.849 0.128 -1.94 0.883 0.103 -2.39 Movera 0.0615 0.0577 3.65 0.111 0.0985 2.48 0.0615 0.0577 3.65 Start date (reg.) 17 835 858 335 0.0377 18 432 323 954 0.117 17 835 858 361 -0.07 Duration (reg.) 1298 611 174 1.31 1038 288 864 0.958 1298 611 158 1.4 Registered t-2 0.866 0.116 -2.15 0.817 0.15 -1.64 0.866 0.116 -2.15 Registered t-3 0.72 0.201 -0.983 0.679 0.218 -0.765 0.72 0.201 -0.983 Switched PCC (2–4 years pre) 0.622 0.235 -0.505 0.481 0.25 0.0744 0.622 0.235 -0.504 Switched GP (2–4 years pre) 0.703 0.209 -0.89 0.543 0.248 -0.173 0.703 0.209 -0.89 First diagnosis of depression 0.338 0.224 0.683 0.373 0.234 0.523 0.338 0.224 0.683 First diagnosis of hypertension 0.318 0.217 0.783 0.28 0.202 0.978 0.318 0.217 0.783 First diagnosis of ischaemic heart disease 0.0529 0.0501 3.99 0.0419 0.0402 4.57 0.0529 0.0501 4 Index year 2012 4.42 0.0325 2013 2.59 -0.333 2012 4.42 0.0972 Any GP visit (3–4 years pre) 0.522 0.25 -0.0861 0.495 0.25 0.0201 0.521 0.25 -0.086 Number of GP visits (3–4 years pre) 1.33 3.9 2.87 1.23 3.56 2.68 1.33 3.9 2.91 Number of nurse visits (3–4 years pre) 1.3 11.5 7.82 1.29 10.9 7.05 1.3 11.5 8.91 Any ED visit (3–4 years pre) 0.131 0.114 2.19 0.114 0.101 2.44 0.131 0.114 2.19 Any GP visit (1–2 years pre) 0.867 0.115 -2.16 0.884 0.102 -2.4 0.867 0.115 -2.16 Number of GP visits (1–2 years pre) 2.76 7.21 2.31 2.83 7.11 2.16 2.76 7.21 2.36 Number of nurse visits (1–2 years pre) 2.18 18.2 6.44 2.67 27.2 4.11 2.18 18.2 4.79 The table shows the first three sample moments — the mean, the variance, and the skewness — for the background characteristics which were used to obtain entropy weights. For the comparison group with no named GP, A shows the moments as observed in the data, and B shows the moments after using the entropy weights. The start date and duration variables refer to the PCC registration that was ongoing at the index date. The dummy variables for being registered in t-2 and t-3 refer to the second and third year before the index date (by definition, everyone was registered the year before the index date). The dummy variables for having switched PCC or GP indicate whether such a switch occurred during the 2–4 years before the index date. The prior care use variables at the bottom of the table are calculated over the 3–4 and 1–2 years before the index date, respectively.
aAn individual who recently moved from another municipality.
ED = emergency department. Disp. inc = disposable income. PCC = primary care centre. Pre = period before the index date. Reg = registration.
Named GP No named GP Obs Mean SD Obs Mean SD Continuity UPC 17 044 0.484 0.236 28 591 0.457 0.217 UPC first 6412 0.675 0.255 11 108 0.638 0.248 Care utilisation Number of GP visits 23 762 5.95 5.29 42 301 5.32 5.04 Number of GP visits for first chronic condition 23 762 1.83 2.52 42 301 1.73 2.49 Number of nurse visits 23 762 5.18 8.41 42 301 5.01 8.89 Any OOH visit 23 762 0.175 0.38 42 301 0.19 0.392 Any ED visits 23 762 0.354 0.478 42 301 0.348 0.476 Any unplanned hospitalisation 23 762 0.233 0.423 42 301 0.226 0.418 Attrition Still registered t+1 23 762 0.974 0.16 42 301 0.968 0.177 Still registered t+2 23 762 0.956 0.204 42 301 0.944 0.23 Still registered t+3 23 762 0.94 0.237 42 301 0.924 0.265 Still registered t+4 23 762 0.925 0.263 42 301 0.907 0.291 Death (1 year) 23 762 0.0152 0.122 42 301 0.0134 0.115 Death (2 years) 23 762 0.0241 0.153 42 301 0.0208 0.143 Death (3 years) 23 762 0.0335 0.18 42 301 0.0289 0.167 Death (4 years) 20 703 0.0454 0.208 35 511 0.0396 0.195 The table shows the mean (or proportion for binary variables) and standard deviation (SD) of the outcome variables for the sample used in each model (Obs). The sample generally includes the whole study population, with the following exceptions: 1) For the two continuity variables (usual provider of care [UPC] index), the sample includes individuals with at least three GP visits during follow-up; for visits with any diagnosis (UPC) and for visits related to the new chronic condition (UPC first). 2) For the variable indicating mortality in t+4, the sample excludes patients diagnosed in 2015, because of a lack of data regarding deaths in the final follow-up year for that cohort.
ED = emergency department. OOH = out of hours.
Any diagnosis Chronic diagnosis 1 2 3 4 5 6 Named GP 0.0276a
(0.0139)0.0264
(0.0169)-0.000322
(0.00756)0.0379a
(0.0177)0.0330
(0.0299)0.0185
(0.0135)n 45 635 45 635 45 635 17 520 17 520 17 520 Entropy balancing weights No Yes Yes No Yes Yes PCC FE No No Yes No No Yes The table shows the estimated association (regression coefficients) between having a named GP and continuity of care for three model specifications. Model 1 (columns 1 and 4) shows the unadjusted difference in means. Model 2 (columns 2 and 5) applies entropy balancing weights. Model 3 (columns 3 and 6) applies entropy balancing weights and includes practice fixed effects (PCC FE). Continuity is measured by the usual provider of care (UPC) index for patients with at least three GP visits during follow-up. Standard errors (in brackets) are clustered by PCC.
a P<0.05
1 2 3 4 5 6 A Number of GP visits (any diagnosis) Number of GP visits (chronic diagnosis) Named GP 0.628c 0.411b 0.364c 0.104 0.0642 -0.0112 (0.130) (0.143) (0.0915) (0.0590) (0.0759) (0.0562) B Number of nurse visits Any OOH visit Named GP 0.164 -0.326 0.104 -0.0147 0.00944 0.00605 (0.198) (0.271) (0.183) (0.0109) (0.0129) (0.00599) C Any ED visit Any hospitalisation Named GP 0.00630 0.0438a 0.0128 0.00708 -0.0106 0.00149 (0.0143) (0.0215) (0.00854) (0.00555) (0.0108) (0.00704) n 66 063 66 063 66 063 66 063 66 063 66 063 EBW No Yes Yes No Yes Yes PCC FE No No Yes No No Yes The table shows the estimated associations between having a named GP and the secondary outcomes. Model 1 (columns 1 and 4) shows the unadjusted difference in means. Model 2 (columns 2 and 5) applies entropy balancing weights. Model 3 (columns 3 and 6) applies entropy balancing weights (EBW) and includes practice fixed effects (PCC FE). Standard errors (in brackets) are clustered by PCC. * P<0.05, ** P<0.01, *** P<0.001.
ED = emergency department. OOH = out-of-hours clinic.
a P<0.05. b P<0.01. c P<0.001.
Supplementary Data
- LME_10.3399BJGPO.2024.0118.pdf -
Supplementary material is not copyedited or typeset, and is published as supplied by the author(s). The author(s) retain(s) responsibility for its accuracy.
In this issue
Jump to section
More in this TOC Section
Related Articles
Cited By...
