Exploring syndemic vulnerability among adolescents living in urban cities in the Netherlands: a latent class analysis
Samantha Frederika Francisca Groenestein, Sabine Plag, Eveline Margaretha Dubbeldeman, Robert R J M Vermeiren, Jet Bussemaker, Suzan van der Pas, Matty R Crone
TL;DR
This study identifies clusters of health conditions among Dutch adolescents and links them to social factors and higher healthcare use.
Contribution
The paper introduces a novel application of syndemic theory to understand adolescent multimorbidity in high-income urban settings.
Findings
25.7% of adolescents had at least two health conditions, forming four distinct multimorbidity classes.
Classes with behavioral and internalizing conditions were linked to adverse social factors like early school dropout.
All multimorbidity classes showed increased healthcare use and costs compared to those with one or no conditions.
Abstract
The increase of adolescent multimorbidity in welfare states is a major concern for current and future population health. However, current health interventions remain insufficient. Syndemic research among adolescents in high-income countries could improve understanding of mechanisms contributing to social and health inequalities. This study explores the existence of clustered health conditions among adolescents aged 10–19 registered at general practices in two average Dutch cities: The Hague and Leiden. We examine which social and contextual factors are associated with these clustered health conditions and to what extent these clusters relate to healthcare use and costs. This cross-sectional study used general practitioner registration data on diagnoses from the Extramural Leiden University Medical Centrum Academic Network database to explore multimorbidity based on the 2-year…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
| No morbidity | Single morbidity | Multimorbidity | Total study population | P value | |
|---|---|---|---|---|---|
| n=4780 (44.1%) | n=3280 (30.3%) | n=2781 (25.7%) | n=10 841 (100%) | ||
| n (%) | n (%) | n (%) | n (%) | ||
| Age | |||||
| 10–14 years | 2797 (58.5) | 1795 (54.7) | 1362 (49.0) | 5954 (54.9) | <0.001 |
| 15–19 years | 1983 (41.5) | 1485 (45.3) | 1419 (51.0) | 4887 (45.1) | |
| Sex | |||||
| Male | 2600 (54.4) | 1601 (48.8) | 1274 (45.8) | 5475 (50.5) | <0.001 |
| Female | 2180 (45.6) | 1679 (51.2) | 1507 (54.2) | 5366 (49.5) | |
| Country of origin | |||||
| The Netherlands | 2113 (44.2) | 1503 (45.8) | 1268 (45.6) | 4884 (45.1) | <0.001 |
| Europe (excl. The Netherlands) | 427 (8.9) | 240 (7.3) | 192 (6.9) | 859 (7.9) | |
| Turkey/Morocco | 971 (20.3) | 765 (23.3) | 653 (23.5) | 2389 (22.0) | |
| Suriname/Indonesia/Dutch Caribbean | 514 (10.8) | 352 (10.7) | 314 (11.3) | 1180 (10.9) | |
| Other (Africa/Asia/America/Oceania) | 755 (15.8) | 420 (12.8) | 354 (12.7) | 1529 (14.1) | |
| Juvenile crime suspect | |||||
| No | 4633 (96.9) | 3147 (95.9) | 2677 (96.3) | 10 607 (97.8) | <0.01 |
| Yes | 147 (3.1) | 133 (4.1) | 132 (4.7) | 412 (3.8) | |
| School dropout | |||||
| No | 4588 (96.0) | 3126 (95.3) | 2569 (92.4) | 10 283 (94.9) | <0.001 |
| Yes | 192 (4.0) | 154 (4.7) | 212 (7.6) | 558 (5.1) | |
| Victim of a criminal act | |||||
| No | 4491 (94.0) | 3007 (91.7) | 2459 (88.4) | 9957 (91.8) | <0.001 |
| Yes | 289 (6.0) | 273 (8.3) | 322 (11.6) | 884 (8.3) | |
| Social problems | |||||
| No | 4597 (96.2) | 3101 (94.5) | 2549 (91.7) | 10 247 (94.5) | <0.001 |
| Yes | 183 (3.8) | 179 (5.5) | 232 (8.3) | 594 (5.5) | |
| Single parent household | |||||
| No | 3607 (75.5) | 2469 (75.3) | 1949 (70.1) | 8025 (74.0) | <0.001 |
| Yes | 1173 (24.5) | 811 (24.7) | 832 (29.9) | 2816 (26.0) | |
| Household composition | |||||
| 4 or less | 2963 (62.0) | 2144 (65.4) | 1948 (70.0) | 7055 (65.1) | <0.001 |
| More than 4 | 1817 (38.0) | 1136 (34.6) | 833 (30.0) | 3786 (34.9) | |
| Known at youth protection services | |||||
| No | 4714 (98.6) | 3216 (98.0) | 2677 (96.3) | 10 607 (97.8) | |
| Yes | 66 (1.4) | 64 (2.0) | 104 (3.7) | 234 (2.2) | <0.001 |
| Household income | |||||
| Low | 702 (14.7) | 445 (13.6) | 436 (15.7) | 1583 (14.6) | <0.001 |
| Moderate | 3177 (66.5) | 2394 (73.0) | 2005 (72.1) | 7576 (69.9) | |
| High | 786 (16.4) | 390 (11.9) | 270 (9.7) | 1446 (13.3) | |
| Other | 115 (2.4) | 51 (1.6) | 70 (2.5) | 236 (2.2) | |
| Socioeconomic status neighbourhood | |||||
| Low | 2539 (53.1) | 1766 (53.8) | 1473 (53.0) | 5778 (53.3) | <0.001 |
| Average | 446 (9.3) | 248 (7.6) | 192 (6.9) | 886 (8.2) | |
| High | 1794 (37.5) | 1266 (38.6) | 1116 (40.1) | 4176 (38.5) | |
| Parent(s) having debts | |||||
| No | 4323 (90.4) | 2958 (90.2) | 2438 (87.7) | 9719 (89.7) | −0.001 |
| Yes | 457 (9.6) | 322 (9.8) | 343 (12.3) | 1122 (10.3) | |
| Parent(s) being detained or suspect of a crime | |||||
| No | 4421 (92.5) | 2993 (91.3) | 2501 (89.9) | 9915 (91.5) | 0.002 |
| Yes | 359 (7.5) | 287 (8.8) | 280 (10.1) | 926 (8.5) | |
| Parent(s) being victim of a criminal act | |||||
| No | 3867 (80.9) | 2599 (79.2) | 2186 (78.6) | 8652 (79.8) | 0.082 |
| Yes | 913 (19.1) | 681 (20.8) | 595 (21.4) | 2189 (20.2) | |
| Parent(s) being divorced | |||||
| No | 4274 (89.4) | 2906 (88.6) | 2482 (89.2) | 9662 (89.1) | 0.706 |
| Yes | 506 (10.6) | 374 (11.4) | 299 (10.8) | 1179 (10.9) | |
| Highest achieved educational level parent(s) | |||||
| Low | 1148 (24.0) | 858 (26.2) | 818 (29.4) | 2824 (26.0) | <0.001 |
| Middle | 842 (17.6) | 676 (20.6) | 629 (22.6) | 2147 (19.8) | |
| High | 1985 (41.5) | 1263 (38.5) | 949 (34.1) | 4197 (38.7) | |
| Missing | 805 (16.8) | 483 (14.7) | 385 (13.8) | 1673 (15.4) | |
| Parent(s) with a chronic physical health condition | |||||
| None | 1847 (12.1) | 847 (25.8) | 473 (17.0) | 3167 (29.2) | <0.001 |
| One parent | 2181 (29.8) | 1611 (49.1) | 1441 (51.8) | 5233 (48.3) | |
| Both parents | 752 (28.1) | 822 (25.1) | 867 (31.2) | 2441 (22.5) | |
| Parent(s) with a mental disorder | |||||
| None | 2977 (62.3) | 1544 (47.1) | 1021 (36.7) | 5542 (51.1) | <0.001 |
| One parent | 1525 (31.9) | 1400 (42.7) | 1316 (47.3) | 4241 (39.1) | |
| Both parents | 278 (5.8) | 336 (10.2) | 444 (16.0) | 1058 (9.8) | |
| Parent(s) with a somatic health condition | |||||
| None | 2048 (42.8) | 998 (30.4) | 686 (24.7) | 3732 (34.4) | <0.001 |
| One parent | 2090 (43.7) | 1649 (50.3) | 1465 (52.7) | 5204 (48.0) | |
| Both parents | 642 (13.4) | 633 (19.3) | 630 (22.7) | 1905 (17.6) | |
| Parent(s) with social problems | |||||
| No | 3937 (82.4) | 2490 (75.9) | 1924 (69.2) | 8351 (77.0) | <0.001 |
| Yes | 843 (17.6) | 790 (24.1) | 857 (30.8) | 2490 (23.0) | |
| GP healthcare use | |||||
| No visits | 2953 (61.8) | 1004 (30.6) | 353 (12.7) | 4310 (39.8) | <0.001 |
| 1–3 visits | 1301 (27.2) | 1229 (37.5) | 754 (27.1) | 3284 (30.3) | |
| 4–6 visits | 388 (8.1) | 654 (19.9) | 699 (25.1) | 1741 (16.1) | |
| More than 6 visits | 138 (2.9) | 393 (12.0) | 975 (35.1) | 1506 (13.9) | |
| Healthcare costs | |||||
| Through 385 | 2438 (51.0) | 1236 (37.7) | 642 (23.1) | 4316 (39.8) | <0.001 |
| 385–2000 | 2097 (43.9) | 1769 (53.9) | 1698 (61.1) | 5564 (51.3) | |
| 2000 or more | 213 (4.5) | 269 (8.2) | 438 (15.7) | 920 (8.5) | |
| Missing | 32 (0.7) | 6 (0.2) | 3 (0.1) | 41 (0.4) | |
| In contact with youth care | |||||
| No | 3707 (77.6) | 2223 (67.8) | 1424 (51.2) | 7354 (67.8) | <0.001 |
| Yes | 1073 (22.4) | 1057 (32.2) | 1357 (48.8) | 3487 (32.2) | |
| Single morbidity | Multimorbidity | Total study population | |
|---|---|---|---|
| n=3280 | n=2781 | n=10 841 | |
| n (%) | n (%) | n (%) | |
| Endocrine disorders | 27 (0.8) | 71 (2.6) | 98 (0.9) |
| Cancer | 11 (0.3) | 29 (1.0) | 40 (0.4) |
| Cardiac disease | 13 (0.4) | 22 (0.8) | 35 (0.3) |
| (Chronic) neck and back conditions | 253 (7.7) | 840 (7.7) | |
| Headache disorders | 182 (5.5) | 763 (7.0) | |
| Asthma | 259 (7.9) | 867 (8.0) | |
| Gastrointestinal conditions | 367 (11.2) | 1208 (11.1) | |
| Generalised fatigue/pain | 291 (8.9) | 942 (8.7) | |
| Skin conditions | 797 (24.3) | 2056 (19.0) | |
| Overweight/obesity | 106 (3.2) | 450 (4.2) | |
| (Head) trauma and fractures | 209 (6.4) | 663 (6.1) | |
| Epilepsy | 23 (0.7) | 53 (1.9) | 76 (0.7) |
| Sexual behaviour | 30 (0.9) | 50 (1.8) | 80 (0.7) |
| Externalising behavioural disorders | 300 (9.1) | 874 (8.1) | |
| Internalising disorders | 241 (7.3) | 848 (7.8) | |
| Burnout/stress | 116 (3.5) | 413 (3.8) | |
| Suicide attempts | 20 (0.7) | ||
| Substance abuse | 31 (0.9) | 78 (2.8) | 109 (1.0) |
| Other mental disorders | 19 (0.6) | 95 (3.4) | 114 (1.1) |
| Class 1 | Class 2 | Class 3 | Class 4 | |
|---|---|---|---|---|
| Asthma-skin conditions | Skin conditions-pain | Externalising behavioural disorders and skin conditions | Gastrointestinal conditions, internalising disorders and pain | |
| Class membership probabilities (%) | 14.7 | 29.2 | 18.2 | 37.9 |
| Probability prevalence based on the LCA | 15.9 | 28.6 | 20.3 | 35.2 |
| (Chronic) neck and back conditions | 11.9 |
| 13.3 |
|
| Headache disorders | 14.5 |
| 13.1 |
|
| Asthma |
| 0 |
| 8.8 |
| Gastrointestinal conditions |
|
| 18.7 |
|
| Generalised fatigue/pain | 10.6 |
| 17.1 |
|
| Skin conditions |
|
|
| 0 |
| Overweight/obesity | 10.5 | 12.1 | 8.9 | 15.0 |
| (Head) trauma and fractures | 8.9 | 13.4 | 16.6 |
|
| Externalising behavioural disorders | 0 | 2.3 |
| 4.6 |
| Internalising disorders | 11.7 | 19.0 | 12.9 |
|
| Burnout/stress | 6.3 | 10.3 | 9.0 | 13.5 |
| Single morbidity | Class 1 | Class 2 | Class 3 | Class 4 | |
|---|---|---|---|---|---|
| n=3280 (30.3%) | n=442 (4.1%) | n=794 (7.3%) | n=565 (5.2%) | n=980 (9.0%) | |
| OR (95% CI) | OR (95%CI) | OR (95% CI) | OR (95% CI) | OR (95% CI) | |
| Intercept | |||||
| Age | |||||
| 10–14 years | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| 15–19 years | 1.09 (0.98 to 1.21) | 1.16 (0.93 to 1.46) | 1.04 (0.88 to 1.25) | 1.81 (1.54 to 2.14) | |
| Sex | |||||
| Male | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Female | 1.11 (0.90 to 1.38) | 2.08 (1.78 to 2.44) | |||
| Country of origin | |||||
| The Netherlands | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Europe (excl. The Netherlands) | 0.92 (0.75 to 1.13) | 0.84 (0.52 to 1.37) | 1.10 (0.77 to 1.57) | 0.87 (0.58 to 1.31) | 1.16 (0.84 to 1.58) |
| Turkey/Morocco | 1.01 (0.86 to 1.18) | 1.17 (0.85 to 1.62) | 1.13 (0.88 to 1.47) | 1.10 (0.87 to 1.40) | |
| Suriname/Indonesia/Dutch-Caribbean | 0.95 (0.80 to 1.12) | 0.90 (0.62 to 1.32) | 1.08 (0.81 to 1.44) | 0.77 (0.56 to 1.06) | 0.89 (0.68 to 1.17) |
| Other (Africa/Asia/America/Oceania) | 0.87 (0.74 to 1.03) | 1.07 (0.76 to 1.50) | 0.88 (0.66 to 1.17) | 0.83 (0.61 to 1.14) | 0.90 (0.70 to 1.17) |
| Juvenile crime suspect | |||||
| No | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 1.29 (0.98 to 1.70) | 1.34 (0.78 to 2.31) | 0.77 (0.45 to 1.32) | 1.26 (0.86 to 1.84) | |
| School dropout | |||||
| No | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 1.01 (0.78 to 1.30) | 1.13 (0.68 to 1.86) | 1.03 (0.67 to 1.57) | ||
| Victim of a criminal act | |||||
| No | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 0.86 (0.55 to 1.35) | ||||
| Social problems | |||||
| No | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 1.13 (0.89 to 1.44) | 1.44 (0.92 to 2.25) | |||
| Single parent household | |||||
| No† | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 0.96 (0.84 to 1.10) | 1.09 (0.87 to 1.35) | 1.13 (0.92 to 1.38) | ||
| Household composition | |||||
| 4 or less | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| More than 4 | 0.85 (0.67 to 1.10) | ||||
| Known at Youth Protection Services | |||||
| No | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 1.35 (0.92 to 1.99) | 0.73 (0.28 to 1.93) | 1.37 (0.74 to 2.53) | ||
| Household income | |||||
| Low | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Moderate | 1.20 (0.88 to 1.64) | 1.06 (0.83 to 1.35) | 1.06 (0.79 to 1.41) | 1.21 (0.96 to 1.52) | |
| High | 0.88 (0.71 to 1.10) | 0.92 (0.57 to 1.49) | 0.74 (0.48 to 1.14) | 1.04 (0.74 to 1.47) | |
| Other | 0.92 (0.59 to 1.45) | 1.87 (0.82 to 4.25) | 0.58 (0.24 to 1.43) | 1.43 (0.81 to 2.52) | |
| SES neighbourhood | |||||
| Low | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Average | 0.88 (0.73 to 1.07) | 1.19 (0.80 to 1.77) | 0.84 (0.59 to 1.20) | 0.83 (0.55 to 1.24) | 0.81 (0.58 to 1.12) |
| High | 1.13 (1.00 to 1.28) | 1.23 (0.94 to 1.60) | |||
| Parent(s) having debts | |||||
| No | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 0.95 (0.80 to 1.13) | 0.87 (0.60 to 1.25) | 0.92 (0.69 to 1.22) | 1.15 (0.85 to 1.54) | 1.10 (0.85 to 1.41) |
| Parent(s) being detained or suspect of a crime | |||||
| No | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 1.00 (0.83 to 1.21) | 0.82 (0.54 to 1.24) | 0.91 (0.67 to 1.24) | 1.26 (0.92 to 1.72) | 0.94 (0.72 to 1.25) |
| Parent(s) being victim of a criminal act | |||||
| No | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 1.00 (0.88 to 1.13) | 0.80 (0.61 to 1.06) | 0.90 (0.73 to 1.11) | 1.02 (0.81 to 1.30) | 0.99 (0.81 to 1.20) |
| Parent(s) being divorced | |||||
| No | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | 0.98 (0.83 to 1.16) | 0.71 (0.54 to 0.92) | |||
| Highest achieved educational level parent(s) | |||||
| Low | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Middle | 1.06 (0.92 to 1.23) | 1.15 (0.86 to 1.54) | 1.04 (0.83 to 1.30) | 1.27 (0.98 to 1.66) | 0.94 (0.76 to 1.17) |
| High | 0.98 (0.85 to 1.13) | 0.91 (0.68 to 1.23) | 0.81 (0.64 to 1.02) | 1.00 (0.76 to 1.31) | |
| Parent(s) with a chronic physical health condition | |||||
| None | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| One parent | |||||
| Both parents | |||||
| Parent(s) with a mental disorder | |||||
| None | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| One parent | |||||
| Both parents | |||||
| Parent(s) with a somatic health condition | |||||
| None | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| One parent | |||||
| Both parents | |||||
| Parent(s) with social problems | |||||
| No | 1.00 | 1.00 | 1.00 | 1.00 | 1.00 |
| Yes | |||||
| GP healthcare use** | No visits | 1–3 visits | 4–6 visits | More than 6 visits | ||||
|---|---|---|---|---|---|---|---|---|
| Class memberships | n (%) | OR (95% CI) | n (%) | OR (95% CI) | n (%) | OR (95% CI) | n (%) | OR (95% CI) |
| Intercept | 0.44(0.41 to 0.47) | 0.13(0.12 to 0.15) | 0.05(0.04 to 0.06) | |||||
| No morbidity | 2953 (61.8) | 1.00 | 1301 (27.2) | 1.00 | 388 (8.1) | 1.00 | 138 (2.9) | 1.00 |
| Single morbidity | 1004 (30.6) | 1.00 | 1229 (37.5) | 2.78(2.50 to 3.09) | 654 (19.9) | 4.96(4.29 to 5.73) | 393 (12.0) | 8.38(6.81 to 10.30) |
| Class 1 | 47 (10.6) | 1.00 | 135 (30.5 | 6.52(4.65 to 9.14) | 116 (25.3) | 18.78(13.17 to 26.79) | 144 (32.6) | 65.56(45.25 to 95.00) |
| Class 2 | 77 | 1.00 | 220 (27.7) | 6.49(4.96 to 8.48) | 198 (24.9) | 19.57(14.74 to 25.99) | 299 | 83.10(61.37 to 112.52) |
| Class 3 | 58 (10.3) | 1.00 | 176 (31.2) | 6.89(5.08 to 9.33) | 156 (27.6) | 20.47(14.87 to 28.18) | 175 (31.0) | 64.57(45.84 to 90.95) |
| Class 4 | 171 (17.4) | 1.00 | 223 (22.8) | 2.96(2.40 to 3.65) | 229 (23.4) | 10.19(8.14 to 12.76) | 357 (36.4) | 44.67(34.80 to 57.35) |
|
|
|
|
|
| ||||
|
|
|
|
|
|
|
|
| |
| Intercept | 0.86 (0.81 to 0.91) | 0.09 (0.08 to 0.10) | ||||||
| No morbidity | 2438 (51.0) | 1.00 | 2097 (43.9) | 1 | 213 (4.5) | 1 | 32 (0.7) | |
| Single morbidity | 1236 (37.7) | 1.00 | 1769 (53.9) | 1.66 (1.52 to 1.83) | 269 (8.2) | 2.49 (2.06 to 3.02) | 6 (0.2) | |
| Class 1 | 117 (26.5) | 1.00 | 263 (59.5) | 2.61 (2.09 to 3.27) | 61 (13.8) | 5.97 (4.25 to 8.38) | 32 (0.7) | |
| Class 2 | 186 (23.4) | 1.00 | 486 (61.2) | 3.04 (2.54 to 3.63) | 121 (15.2) | 7.45 (5.69 to 9.74) | 1 (0.1) | |
| Class 3 | 132 (23.4) | 1.00 | 356 (63.0) | 3.14 (2.55 to 3.86) | 77 (13.6) | 6.68 (4.88 to 9.14) | 0 (0) | |
| Class 4 | 207 (21.1) | 1.00 | 593 (60.5) | 3.33 (2.81 to 3.94) | 179 (18.3) | 9.90 (7.75 to 12.64) | 1 (0.1) | |
|
|
|
| ||||||
|
|
|
|
|
| ||||
| Intercept | 0.29 (0.27 to 0.31) | |||||||
| No morbidity | 3707 (77.6) | 1.00 | 1073 (22.4) | 1.00 | ||||
| Single morbidity | 2223 (67.8) | 1.00 | 1057 (32.2) | 1.64 (1.49 to 1.82) | ||||
| Class 1 | 306 (67.8) | 1.00 | 136 (30.8) | 1.54 (1.24 to 1.90) | ||||
| Class 2 | 485 (61.1) | 1.00 | 309 (38.9) | 2.20 (1.88 to 2.58) | ||||
| Class 3 | 113 (20.0) | 1.00 | 452 (80.0) | 13.82 (11.12 to 17.17) | ||||
| Class 4 | 520 (53.1) | 1.00 | 460 (46.9) | 3.06 (2.65 to 3.52) | ||||
- —NWO/ZonMw
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsChronic Disease Management Strategies · Health disparities and outcomes · Adolescent and Pediatric Healthcare
Introduction
The ongoing increase in multimorbidity, defined as the co-occurrence of two or more health conditions, is a major concern for current and future population health. While multimorbidity is mostly linked to ageing populations, it also adversely affects younger groups’ health, well-being and social participation. Moreover, some chronic health conditions that emerge during adolescence persist into adulthood, contributing to adverse health outcomes and reinforcing social inequalities later in life.1 2
Shared exposure to a set of adverse social conditions creates the circumstances for health conditions to cluster and, consequently, increases health inequalities among individuals and populations.35 This phenomenon is called a syndemic, which postulates that contextual conditions encourage the clustering of health conditions. When two or more diseases or health conditions cluster, they may interact, leading to more adverse health outcomes in individuals and populations than would be expected from having just one condition.610 Nevertheless, current healthcare systems and interventions focusing on adolescent health are designed to treat single diseases or health conditions, putting adolescents with multimorbidity or syndemics at risk of inefficient, contradictory or duplicative services.1 11
Although syndemic research in adolescent populations is limited, some studies have described multimorbidity and the effects of social contextual factors on adolescents’ health. Mental–physical multimorbidity is often described as a prevalent issue in the adolescent population.1215 For example, anxiety, mood disorders and attention-deficit/hyperactivity disorder (e.g., ADHD) tend to co-occur with asthma, obesity, allergies and epilepsy.14
Social contextual factors, operating on the individual, household and parental levels, may act as risk and/or protective factors for the occurrence of health conditions, explaining why some individuals are more vulnerable to adverse health outcomes than others.3 16 For example, research shows that early exposure to parental stress related to finances, work or income is associated with poor health outcomes or psychological distress later in life.5 17 18 Additionally, parental physical health predominantly affects the overall health of children aged 10–12, while parental mental health is a more important factor for children aged 13–15.19
A unique syndemic study in Western Europe by Slagboom et al. provides narrative evidence for the presence of syndemic vulnerability within families and across generations, suggesting that syndemic vulnerability might manifest at a young age.5 However, adolescent multimorbidity studies and syndemic research often focus on adults or specific (clinical) populations or lack generalisability due to a small age range and a limited set of included health conditions.2027 Additionally, much of the syndemic research has been conducted in developing, low-income or middle-income countries.2831
Despite the Netherlands being known as a generous welfare state with organised care, significant social and health inequalities continue to exist between neighbourhoods in urban areas.32 Using a syndemic approach to explore multimorbidity in relation to social contextual factors among adolescents can improve our understanding of the mechanisms that potentially contribute to social and health inequalities. This approach informs an integrated strategy that combines health, social care and preventative activities to address health inequalities at an early age and promote efficient healthcare. Therefore, this study:
Explores the existence of clustered health conditions among adolescents aged 10–19, registered at primary care practices in the Dutch urban cities of The Hague and Leiden.Identifies characteristics at the individual, household and parental level associated with clusters of health conditions.Compares healthcare use and costs between adolescents with clustered health conditions, a single condition and no conditions.
Our findings advance our understanding of adolescents’ vulnerability to syndemics, and whose multimorbidity may be likelier to persist into adulthood, causing lower well-being and worse health.
Methods
This cross-sectional population-based study is part of the project ‘Countering syndemic vulnerability: A community resilience approach’. The project aims to develop a community resilience approach in vulnerable neighbourhoods in the Netherlands, employing syndemic theory. This substudy contributes to mapping the syndemic vulnerability in the participating cities within the project.
For this study, a unique dataset was used, integrating routinely registered primary care records of general practitioners (GPs) connected to the Extramural Leiden University Medical Centrum Academic Network (ELAN) primary care network, and societal data routinely collected by the System of Social Statistical Datasets (SSD) from Statistics Netherlands (SN). ELAN provides medical patient data collected in the primary care setting in the Leiden and The Hague region.33 It includes information on GP-registered disease episodes and the period of concern regarding a health issue, coded according to the WHO International Classification of Primary Care (ICPC). The SSD consists of routinely collected individual-level societal data, including sociodemographic, socioeconomic and household characteristics.34 Data from both sources were coded, anonymised, stored securely within the SN environment and linked through anonymised individual ID numbers.
Study population
The study population comprised all adolescents registered with a participating GP and living in two Dutch cities, The Hague and Leiden. The Hague is one of the largest cities in The Netherlands with more residents with a migration background, a lower average socioeconomic status (SES) score and a smaller prevalence of residents with a good self-perceived health compared with Leiden.35 36 Both cities exhibit disparities in health outcomes and SES scores across neighbourhoods. However, the city of The Hague shows larger differences in SES scores between neighbourhoods.37 Following the WHO, adolescents were defined as being 10–19 years old as of 1 January 2018. In the Netherlands, all residents are registered with a GP, usually in their neighbourhood, making the GP-registered population an adequate representation of the adolescent population living in both cities. Our study population consisted of all adolescents aged 10–19 years who were registered with a participating GP in one of the cities for at least 12 months during 2018–2019 and were alive throughout both years. We determined an individual’s residence in The Hague or Leiden based on their longest registered address in 2018.
Patient and public involvement
No patients or members of the public were involved in formulating the research question, selecting the outcome measures, or in the development of the design or implementation of the study.
Health conditions
The selection of health conditions in our study was based on prevalent health conditions in adolescents, prior research on comorbidity and multimorbidity in adolescents, and the available data.1 11 38 In this study, we use the term ‘health conditions’ to define chronic health conditions or health conditions that have a long-lasting effect. We subdivided health conditions into physical (e.g., asthma, cancer and epilepsy), mental (e.g., externalising behavioural disorders, such as ADHD, and internalising disorders, such as anxiety and feeling depressed) and psychosomatic (e.g., generalised fatigue/pain and headache disorders) health conditions (online supplemental appendix 1).
In the Netherlands, GPs register symptoms and diagnoses of health conditions as episodes using the ICPC. We identified the presence of health conditions through the ICPC codes related to both symptoms and diagnosis. The 2-year prevalence of these conditions was determined based on whether an individual had registered an active, new or mutated episode of the included ICPC codes in either 2018 or 2019. Adolescents were categorised as having none of the included health conditions (no morbidity), having one of the conditions (single morbidity), or having two or more of the conditions (multimorbidity) across the 2-year period in 2018–2019.
Social contextual factors
We included a wide range of social contextual factors at the individual, household and parental levels (online supplemental appendix 2). These factors were selected for their known long-term impact on health and multimorbidity, based on prior research and available data.5 11 17 18 20
Individual-level factors
The SSD includes a diverse set of individual-level data, such as age, sex, country of origin (based on the individual’s and their parent’s birthplace), and whether the individual was a juvenile crime suspect, a victim of a criminal act, or had dropped out of school early between 2015 and 2019. From ELAN, we included information about adolescents having social problems, as registered by a GP in 2018 and 2019.
Household-level factors
At the household level, we extracted data about household composition (e.g., a single parent household and household composition) in 2018 and socioeconomic characteristics, such as household income in 2018 and SES of the neighbourhood where the individual was living in 2019. Information regarding adolescents in contact with Dutch youth protection services due to concerns about an unsafe family environment from 2015 to 2019 was also extracted.
Parental-level factors
To extract information at the parental level, a unique identification code enabled the linkage of adolescents to their parents. We incorporated information about the highest achieved education level attained by the parents, whether parents had debts, had been imprisoned, were victims of a criminal act or got divorced between 2015 and 2019. ELAN provided information on parents’ physical (e.g., cancer, hypertensive heart disease, stroke), mental (eg, burnout, mood disorders, hyperactivity disorders), and somatic (e.g., headache, generalised fatigue/pain) health conditions as well as their social problems between 2016 and 2019.
Healthcare use and costs
Adolescents’ GP healthcare use, healthcare costs and contacts with youth care were included as indications of the severity of the clustered health conditions. Variables related to GP healthcare use were derived from the total number of GP visits made within 2018 and 2019. The GP is the first healthcare provider a patient visits within the primary care system, without the need for a referral. The GP evaluates the severity of the patient’s complaint, provides treatment, prescribes medication and, when necessary, refers the patient to a secondary care specialist, which requires a formal referral. In the Netherlands, GPs function as gatekeepers, meaning that patients must see a GP before accessing specialised care. They operate in accordance with established guidelines for managing health conditions and adhere to professional standards. While these guidelines provide some consistency in care, the frequency of follow-up visits may still vary depending on the individual patient’s needs.
Healthcare costs included all expenses incurred by the care-using adolescent associated with the Health Insurance Act (e.g., GP costs and hospital care costs), excluding youth care costs, and were calculated as the average healthcare costs in euros from 2018 to 2019. The lowest possible maximum personal contribution towards healthcare costs in the Netherlands is €385 per year, which does not apply to children under 18 years of age.
Adolescents’ contact with youth care in 2018 and 2019 was set as a dichotomous variable. ‘Yes’ indicated involvement in one or more forms of assistance or care under the Youth Act, excluding youth protection and youth rehabilitation, such as support for psychological or parenting problems.
Statistical analysis
The data were prepared using IBM SPSS Statistics V.25. Statistical analyses were conducted using RStudio (V.4.2.3). Descriptive statistics were used to describe the characteristics of adolescents with no morbidity, single morbidity and multimorbidity.
Latent class analysis
Latent class analysis (LCA) was conducted using the ‘poLCA’ package to explore different multimorbidity classes in the total adolescent population aged 10–19 with multimorbidity. This technique was chosen for its ability to group individuals without prior assumptions about the number and nature of the latent classes. The findings of Nichols et al. suggest that low-prevalence diseases may introduce noise into the modelling process.39 Consequently, in the present study, only conditions with a prevalence of at least 5% in the multimorbidity population were included in the analysis.
The LCA was conducted without a priori assumptions about the number of latent classes. We assessed model fit based on fit statistics, the interpretability of classes and the models’ entropy for clearer class separation (online supplemental table 1).40 The prevalence of health conditions within each class was derived from the posterior probability of the condition being present and individuals were assigned to the class with the highest predicted class membership (based on posterior probabilities).
Multinomial logistic regressions
Descriptive statistics were used to describe the presence of individual, household and parental characteristics across the latent classes. Multivariable multinomial logistic regression analyses were performed to explore the association between class memberships and individual, household and parental characteristics and to examine the association between memberships and GP healthcare use and costs. Finally, binary logistic regression analyses were performed to examine the association between class membership and contact with youth care.
The significance level for all regression analyses was set at p<0.05.
Results
The total study population consisted of 10 841 adolescents (table 1). About half of the adolescents were aged 10–14 years (54.9%) and were male (50.5%). The majority lived in The Hague (81.1%) and 45.1% were of Dutch origin (table 1). In total, 2781 (25.7%) adolescents were registered as having at least two health conditions during the study period, indicating multimorbidity.
Table 2 shows the 2-year prevalence of health conditions among the single, multimorbidity and total population groups. We included the 11 most frequent health conditions in the multimorbidity group in the LCA analysis, namely (chronic) neck and back conditions (21.1%), headache disorders (20.9%), asthma (21.9%), gastrointestinal conditions (30.2%), generalised fatigue/pain (23.4%), skin conditions (45.3%), overweight/obesity (12.4%), (head) trauma and fractures (16.3%), externalising behavioural disorders (including ADHD) (20.6%), internalising disorders (21.8%) and burnout/stress (10.7%).
LCA class interpretation
We selected the four-class model, considering the fit indices and interpretability, with an entropy of 0.91 (online supplemental table 1). Although the Akaike information criterion (AIC) continued to decrease slightly as the number of classes was increased, suggesting a more significant number of classes as an optimal fit, the AIC tends to overfit complex models.40
Table 3 presents class proportions and the probability prevalence of health conditions within each class (online supplemental figure 1). Classes were labelled based on the health conditions with the highest probability prevalence within each class and those with striking prevalence differences compared with the other classes and the total study population. The four classes were:
Class 1: ‘asthma–skin conditions’ (n=442, 15.9% in multimorbidity group, 4.1% in total study population)
The first multimorbidity class had the lowest prevalence of adolescents, with 4.1% in the total study population. This class was characterised by the highest likelihood of children having asthma, combined with a high likelihood of skin conditions and gastrointestinal conditions.
Class 2: ‘skin conditions–pain’ (n=794, 28.6% in multimorbidity group, 7.3% in total study population)
The second multimorbidity class had a prevalence of 7.3% in the total study population and was characterised by the highest likelihood of skin conditions. Additional conditions with a high probability prevalence were gastrointestinal conditions, headache disorders, generalised fatigue/pain, and (chronic) neck and back conditions.
Class 3: ‘externalising behavioural disorders and skin conditions’ (n=565, 20.3% in multimorbidity group, 5.2% in total study population)
Of all children included in this study, a prevalence of 5.2% was categorised into the third multimorbidity class. Externalising behavioural disorders were the primary health condition in this class having the highest probability prevalence, followed by a high likelihood of skin conditions and asthma.
Class 4: ‘gastrointestinal conditions, internalising disorders and pain’ (n=980, 35.2% in multimorbidity group, 9.0% in study population)
The fourth class had the highest class prevalence of 9.0% in the total study population. Unlike other classes, class 4 does not distinguish one specific health condition. However, compared with other classes, gastrointestinal conditions, generalised fatigue/pain, internalising disorders, (chronic) neck and back conditions, headache disorders and (head) trauma and fractures had the highest likelihood among adolescents in class 4.
Social contextual factors
Table 4 shows the associations (OR(95% CI)) between social contextual factors and the multimorbidity classes based on the multinomial logistic regression with the no morbidity group as the reference category.
Individual-level factors
On the individual level, female adolescents had higher odds of belonging to class 2 (1.84 (95% CI 1.55 to 2.19)) and lower odds of belonging to class 3 (0.79 (95% CI 0.65 to 0.97)), compared with the no morbidity group. Looking at age, 15–19 years had lower odds of belonging to class 3 (0.80 (95% CI 0.64 to 0.98)). In addition, adolescents being born in Turkey or Morocco or with one or both parents or being born in these countries were less likely to belong to class 3 (0.56 (95% CI 0.40 to 0.77)).
Juvenile crime suspects were likelier to belong to class 3 (1.55 (95% CI 1.01 to 2.38)), and early school dropouts had higher odds of belonging to classes 3 (1.79 (95% CI 1.21 to 2.64)) or 4 (1.56 (95% CI 1.15 to 2.13)). Both victims of a known criminal act and adolescents with GP-registered social problems had higher odds of belonging to classes 2 (1.80 (95% CI 1.34 to 2.41)), 3 (2.08 (95% CI 1.53 to 2.83)) and 4 (1.72 (1.33 to 2.23)).
Household-level factors
Adolescents living in a single parent household had higher odds of belonging to classes 1 (1.37 (95% CI 1.04 to 1.81)) and 3 (1.52 (95% CI 1.19 to 1.93)), adolescents living in a household with more than four persons had lower odds of belonging to classes 2 (0.73 (95% CI 0.60 to 0.90)), 3 (0.65 (95% CI 0.51 to 0.82)) and 4 (0.70 (95% CI 0.58 to 0.84)). Adolescents who had contact with Dutch Youth Protective Services had higher odds of belonging to classes 3 (1.90 (95% CI 1.12 to 3.22)) and 4 (2.43 (95% CI 1.54 to 3.82)).
Adolescents with a high household income were less likely to belong to class 2 (0.71 (95% CI 0.48 to 1.03)), while adolescents with a household income defined as other, referring to individuals with an unknown household income or belonging to a student or institutionalised household, were likelier to belong to class 3 (2.00 (95% CI 1.07 to 3.76)). Adolescents living in a high-SES neighbourhood had higher odds of belonging to classes 2 (1.33 (1.08 to 1.63)), 3 (1.79 (1.42 to 2.26)) and 4 (1.35 (1.11 to 1.63)).
Parental-level factors
Compared with the no morbidity group, adolescents of parents with a higher education level had lower odds of belonging to class 4 (0.78 (95% CI 0.63 to 0.96)). All classes were less likely to have parents divorced in the past 5 years.
When adolescents had one or both parents with a GP-registered physical, mental, somatic health conditions, or with GP-registered social problems, odds on belonging to the group with one morbidity and to all four multimorbidity classes were higher compared with the no morbidity group.
Healthcare use and costs
Table 5 shows both the prevalence of healthcare use and costs of all class memberships and the associations between the multimorbidity classes and healthcare use and costs.
GP healthcare use
Compared with the no and single morbidity groups, all multimorbidity classes had higher odds of having more than 6 GP visits. Class 2 had the highest prevalence (37.7%) followed by class 4 (36.4%).
Healthcare costs
Compared with the no and single morbidity group, all the multimorbidity classes had higher odds of having healthcare costs of 2000 or more euros. Class 4 (gastrointestinal conditions, internalising disorders and pain) had the highest prevalence (18.3%), followed by class 2 (15.2%).
Youth care
Classes 2 (2.20 (95% CI 1.88 to 2.58)), 3 (13.82 (95% CI 11.12 to 17.17)) and 4 (3.06 (95% CI 2.65 to 3.52)), had higher odds of being in contact with youth care between 2015 and 2019, in comparison with the no and single morbidity group. Class 3 had the highest prevalence with more than half of the adolescents in this class having had contact with youth care.
Discussion
The objective of this study was to explore whether, and if so, which clusters of health conditions are prevalent among adolescents aged 10–19 registered at primary care practices in the Dutch cities The Hague and Leiden. Additionally, we aimed to identify contributing social contextual factors and the association of multimorbidity classes with healthcare use and costs. One-quarter of the adolescents were registered as having at least two or more chronic health conditions in 2018 and 2019. To explore clusters of health conditions, these adolescents with multimorbidity were grouped based on probability of each individual fitting into a specific group using LCA. This method allowed for the identification of four distinct multimorbidity classes of health conditions, respectively class 1 (asthma–skin conditions), class 2 (skin conditions–pain), class 3 (externalising behavioural disorders and skin conditions) and, most prevalent, class 4 (gastrointestinal conditions, internalising disorders and pain). Skin conditions were a prevalent diagnosis in three of the classes. The fourth cluster of health conditions did not include skin disease and seemed to have a more psychosomatic nature.
Increased healthcare use and costs across all classes indicate a higher disease burden among adolescents with clustered health conditions, which is consistent with previous literature.41 Adolescents with gastrointestinal conditions, internalising disorders and pain (class 4) had the highest healthcare costs and use. Those with externalising behavioural disorders and skin conditions (class 3), however, most frequently used youth care. These two multimorbidity classes, 3 and 4, were characterised by several more prominent adverse social contextual factors.
The co-occurrence of physical and mental health conditions, along with adverse social conditions, emphasises the relevance of a syndemic approach. Syndemic theory posits that two or more clustered diseases or health conditions can interact within adverse social contextual conditions, leading to more adverse health outcomes.610 The disease clustering of both externalising and internalising disorders with other (somatic) health conditions is correlated with higher healthcare use and costs compared with adolescents having only one health condition. Moreover, these clustered conditions co-occur with pronounced adverse social contextual factors, such as an increased likelihood of school dropout.
Importantly, addressing adolescents’ syndemic vulnerability may require an intergenerational syndemic approach, as clustered health and social conditions were found to be passed down through generations. This demonstrates that syndemic vulnerability is potentially intergenerational.5 Our study showed that a large majority of the adolescents with multimorbidity also had one or both parents with physical, mental and/or somatic health conditions. These findings raise questions about the genetic, biological, social and/or behavioural transmittance of health conditions within families.
Evidence indicates intergenerational correlations or transmission of physical and mental health between parents and their offspring.42 43 Children of parents having a severe disease are at higher risk of developing mental health disorders and of reporting lower self-rated psychosomatic health and life satisfaction, scoring even lower when children themselves had a chronic disease.44 45 Due to the limited number of syndemic studies, our understanding of social contextual risk factors, intergenerational transmission and long-term effects is significantly less comprehensive compared with our knowledge of individual health conditions, highlighting the need for more attention to this clustering of health condition and social contextual circumstances in healthcare. Integrated preventative measures focusing on multi-level aspects are essential to reduce potential syndemic vulnerabilities. One recommended strategy is comprehensive prevention, a broad context-sensitive approach that considers overlapping and interacting adversities.46 A context-sensitive approach, combined with long-term family interventions targeting multiple generations—such as the three-generation approach—can enhance the health of parents, their child and following generations.47 Moreover, a Dutch study, using similar data, showed the early onset of health problems among children aged 0–12 years growing up in adverse or unfavourable social contexts, emphasising the importance of preventive measures from an early age.48
In line with our findings, other adolescent multimorbidity studies show similar multimorbidity classes.154953 Multiple studies have reported on the co-occurrence of asthma and skin conditions.51 52 Skin conditions, externalising behavioural disorders and pain complaints, such as stomach pain and headaches, are also common disease clusters in the literature,49 50 54 as are asthma and externalising disorders.55 56 These clusters of health conditions are often explained as resulting from mutually interacting biological or physical mechanisms.495355 Additionally, the biological mechanism of stress has been described as a potentially important factor in both skin diseases and externalising behavioural disorders.57 58
In line with the literature, the combination of skin diseases and pain was more prevalent among girls, and the combination of skin conditions and externalising behavioural disorders was more prevalent among boys. In our study, the latter group was also younger, less often included adolescents from Moroccan or Turkish origin and more often lived in a high SES neighbourhood. Cultural differences in parents’ recognition of externalising behavioural disorders may explain these findings, as a Dutch study reported lower problem recognition among parents of Moroccan or Turkish origin.59
Adolescents with externalising behavioural disorders were most clearly characterised by their increased use of youth care, emphasising the psychosocial aspect of this multimorbidity. They had an increased likelihood of having dropped out of school, of being a victim of a criminal act or being a juvenile crime suspect, of growing up in a single parent household or in a household that had contact with child protection services, and of having parents with GP-registered social problems. In short, social circumstances are central to this group of adolescents.
Parental stress arising from social circumstances in the family setting may affect the early development of mental health problems.45 60 Children who grow up under parental stress, in an unstable family environment, or experiencing parental divorce are at increased risk of developing problems in internalising or externalising behaviour during adolescence.45 Conversely, parental stress is shown to increase when children show externalising behaviour problems.52 For instance, in children with autism spectrum disorders, parental stress affects internalising and externalising behaviour problems, indicating a parent-driven effect. For externalising behaviours, effects are bidirectional, as the externalising behavioural problems of the child lead to increased parental stress, particularly among fathers.61 Additionally, these social and family problems, which contribute to externalising behavioural disorders, can increase the risks of school dropout and criminal behaviour.6264 For example, adolescents living with a single parent may experience emotional distress when comparing themselves with peers in culturally preferred two-parent families.62
The psychosomatic nature of class 4 combined internalising disorders with gastrointestinal conditions and chronic pains, such as generalised fatigue/pain, neck and back conditions and headache disorders. Studies have shown that internalising disorders can be expressed as psychosomatic pain complaints, such as neck and back pain or headaches.14 Although it is not always clear whether internalising disorders and various pain complaints manifest with a similar aetiology or mutually interact,65 there is some evidence for a bidirectional biological interaction between internalising disorders and gastrointestinal conditions.156668 While an intestinal bacteria imbalance affects one’s mood, this imbalance by itself could be caused by stress, anxiety and depression.6668
The adolescents in class 4 were more often female, older, had less-educated parents and had one of the highest rates of healthcare use and costs and youth care use. Other studies have confirmed this increased healthcare use among adolescents with psychosomatic conditions and internalising disorders.64 Adolescents with internalising behavioural disorders and multimorbidity were more often in contact with family protection services and were likelier to be registered as having dropped out of school and as being victims of a criminal act. In short, in this group of adolescents, social and family complexities seem to drive the clustering of (mental) health problems. Conversely, psychosomatic conditions may increase the risk of social problems, as internalising disorders were found to increase the risk of school dropout.64 The combination of both health and social problems in our data calls for approaches that go beyond treatments targeting individual issues. Consequently, we recommend that healthcare shift towards recognising the diverse perspectives and needs of young patients and strengthening interdisciplinary collaboration between healthcare providers and social support systems.
Strengths and limitations
To the best of our knowledge, this is the first study to explore multimorbidity patterns in adolescents using GP-registered data combined with societal data. This data combination allowed us to include a wide range of ICPC codes, healthcare use and costs, and social contextual factors on the individual, household and parental levels over a longer time period. In addition, using GP-registered health data eliminated the risk of information and recall bias by individuals. Further, in the study of Nichols et al.,39 LCA was found to be superior to other clustering techniques in identifying multimorbidity classes in a simulated dataset where the actual clusters were known.39 Finally, due to the sources of data we utilised the population included in our study is representative for adolescents living in Dutch cities, as it consisted of a large adolescent primary care population with significant differences in health and social well-being.
Our study does have some limitations. First, by using GP registration data, we could not account for potential GPs implicit bias, undiagnosed, incorrectly diagnosed or unreported health conditions, and we cannot draw conclusions about the severity of the registered health conditions. Although we discussed our data with GPs, implicit bias remains an important issue, as it may affect diagnostic and treatment decisions and warrants targeted training. Additionally, the actual number of adolescents with multimorbidity may be higher, as adolescents may not always visit the GP while having symptoms, or may visit a GP outside their own neighbourhood or city. Also, not all GPs in The Hague and Leiden participate in ELAN, so their patients were not captured in our dataset. Second, complete information on social contextual factors was not available for all adolescents; in some cases, one or both parents were not registered in SN or ELAN or their registered information was incomplete, missing factors such as education level. Although information on the country of birth of the individuals and their parents was included, data on immigration status and the duration of residence of families in the Netherlands were not available. Third was that, because we analysed a 2-year period, we did not know if health conditions were present at the same time, and we were not able to identify causal relationships. Similarly, as we did not investigate the interactions between clustered health conditions, we were only able to identify likely syndemic vulnerability rather than definitively determine the presence of a syndemic. Finally, although LCA is an interpretable method to group individuals and the four-class model demonstrated a reliable classification precision with an entropy of 0.91, a limitation arises in assigning adolescents to a class based solely on the largest posterior probability, as not all adolescents fully align with a single class. This should be considered when interpreting the results.
Conclusions
One quarter of adolescents aged 10–19 year-old, registered at primary care practices in the Dutch cities The Hague and Leiden, experienced multimorbidity between 2018 and 2019. Within this group, four multimorbidity classes were identified, shaped by social contextual factors on the individual, household and parental levels. Adolescents in classes 3 and 4, characterised by externalising or internalising disorders with physical and somatic health conditions, most frequently exhibited adverse social contextual factors, such as school dropout. Increased healthcare use and costs across all classes indicate a higher disease burden for adolescents with clustered health conditions compared with those with a single or no condition. This shows the importance of considering the interaction and co-occurrence of health conditions in diagnosis and treatment, and the collaboration with social services.
Our findings highlight the importance of implementing integrated preventative strategies addressing multilevel aspects to reduce potential syndemic vulnerability. The perspectives and needs of young patients and their families should be prioritised, and interdisciplinary collaboration between healthcare providers and social support systems should be promoted. We recommend validating the current LCA model in a comparable study population to clarify whether the identified classes are present in other adolescent primary care population groups. Furthermore, future studies should focus on a longitudinal approach to identify the causal relationships among clustered health conditions and social contextual factors across the life course. Monitoring adolescents’ syndemic vulnerability over time will provide a deeper understanding of health and social inequalities later in life.
Supplementary material
10.1136/bmjph-2024-002032online supplemental file 1
10.1136/bmjph-2024-002032online supplemental file 2
10.1136/bmjph-2024-002032online supplemental file 3
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1van den Akker M Dieckelmann M Hussain MA et al Children and adolescents are not small adults: toward a better understanding of multimorbidity in younger populations J Clin Epidemiol 20221491657110.1016/j.jclinepi.2022.07.00335820585 · doi ↗ · pubmed ↗
- 2Armocida B Monasta L Sawyer S et al Burden of non-communicable diseases among adolescents aged 10-24 years in the EU, 1990-2019: a systematic analysis of the Global Burden of Diseases Study 2019 Lancet Child Adolesc Health 202263678310.1016/S 2352-4642(22)00073-635339209 PMC 9090900 · doi ↗ · pubmed ↗
- 3Sawyer SM Afifi RA Bearinger LH et al Adolescence: a foundation for future health Lancet 201237916304010.1016/S 0140-6736(12)60072-522538178 · doi ↗ · pubmed ↗
- 4National Academies of Sciences E, Medicine, Health, Medicine D, Board on Global H, Forum on Microbial T The national academies collection: reports funded by national institutes of health Minicucci C Using syndemic theory and the societal lens to inform resilient recovery from COVID-19: Toward a post-pandemic World: proceedings of a workshop—in brief Washington (DC)National Academies Press (US)34324281 · pubmed ↗
- 5Slagboom MN Crone MR Reis R Exploring syndemic vulnerability across generations: A case study of a former fishing village in the Netherlands Soc Sci Med 202229511312210.1016/j.socscimed.2020.11312232576403 · doi ↗ · pubmed ↗
- 6Tsai AC Mendenhall E Trostle JA et al Co-occurring epidemics, syndemics, and population health Lancet 20173899788210.1016/S 0140-6736(17)30403-828271848 PMC 5972361 · doi ↗ · pubmed ↗
- 7Mendenhall E Newfield T Tsai AC Syndemic theory, methods, and data Soc Sci Med 202229511465610.1016/j.socscimed.2021.11465634949486 PMC 8669950 · doi ↗ · pubmed ↗
- 8Mendenhall E Kohrt BA Logie CH et al Syndemics and clinical science Nat Med 20222813596210.1038/s 41591-022-01888-y 35864249 · doi ↗ · pubmed ↗