AboutSubmit my research

Spatial Microsimulation, Health

Research articles · Aug 31, 2019

Spatial microsimulation of osteoarthritis prevalence at the small area level in England – Constraint selection for a 2-stage microsimulation process

Onosi Sylvia Ifesemen, Thomas Bestwick-Stevenson, Kimberley L. Edwards

Centre for Sports, Exercise and Osteoarthritis, Versus Arthritis, Nottingham, United Kingdom; Academic Orthopaedics, Trauma and Sports medicine, School of Medicine, Queens Medical Centre, University of Nottingham, Nottingham, United Kingdom

Volume 12(2) Summer 2019

Cite as: O. Ifesemen, T. Bestwick-Stevenson, K. Edwards; 2019; Spatial microsimulation of osteoarthritis prevalence at the small area level in England – Constraint selection for a 2-stage microsimulation process; International Journal of Microsimulation; 12(2); 37–51. doi: 10.34196/ijm.00200

Abstract

The presence of identical benchmark/constraint variables in both geographic and survey datasets is a principal requirement for static spatial microsimulation models, particularly in the field of medicine and health sciences. This is also a key limitation of static spatial models because geographical datasets rarely contain all variables required to realistically simulate an outcome. We believe this challenge can be overcome by a multilevel approach to spatial microsimulation using a case study of estimating the small area level prevalence of knee osteoarthritis in England. In the paper, we describe constraint selection and demonstrate a novel two–stage spatial microsimulation procedure using SimObesity, a static deterministic combinatorial spatial microsimulation model. We also present the validation parameters of our synthetic data, important areas for consideration and avenues for future research. Our findings demonstrate that important benchmark variables absent from the geographical dataset can be incorporated into spatial microsimulation models without compromising model robustness.

This article contains 4 figures and 7 tables.

Figures

Figure 1. 2-stage Spatial Microsimulation
Figure 2. Scatter plots for age, sex and NSSEC categories, simulated counts from the hybrid dataset versus census totals
Figure 3. Scatter plots for age, sex and NSSEC categories, simulated counts from the final dataset versus census
Figure 4. Comparing regional aggregates of our simulated synthetic microdata data with ELSA data (grouped by regions)

Tables

Table 1. Names and categories of covariates used in our analyses
Variable Categories
Age 50–59 years, 60–69 years, 70–79 years, 80–89 years, ≥90 years
Sex Male, Female
*Ethnicity White, Non-White
Health Good, Fair, Poor
BMI Underweight, Overweight & Obese
*NSSEC Higher managerial & professional, Lower-professional, Intermediate Occupations, Lower supervisory & technical, Semi routine, routine & other.
Marital Status Single, Married, Separated, Divorced, Widowed
Level of Education NVQ4&5, Higher Education below Degree, NVQ3, NVQ2, NVQ1
Smoking history Never smoker, Ex-smoker, Current smoker
Alcohol Never, once or twice a year, Every couple of Months, Once or twice a month, Once or twice a week, Every other day, Everyday
Fruit & Vegetable Consumption/week Less than 5 portions, Greater than 5 portions

Categorised to match ELSA.

Table 2. Frequency distribution of variables in geographical (census) and survey (HSE and ELSA) datasets
Variables/categories ELSA, n (%) HSE, n (%) Census LSOA, n (%)
Age 50–59 years 2,062 (22.49) 4,901 (32.67) 6,397,087 (35.09)
60–69 years 3,451 (37.64) 4,018 (32.09) 5,680,431 (31.16)
70–79 years 2,444 (26.66) 2,837(22.66) 3,713,474 (20.37)
80–89 years 1,005 (10.96) 1,383 (11.05) 2,032,084 (11.16)
≥90 years 207 (2.26) 192 (1.53) 403,817 (2.22)
Sex Female 5,096 (55.58) 6,756 (53.96) 9,658,440 (52.98)
Male 4,073 (44.42) 5,765 (46.04) 8,571,453 (47.02)
*Marital Single, Cohabitees 603 (6.58) 1,425 (11.39) 1,464,682 (8.03)
Married, Civil partnerships 5,872 (64.06) 7,726 (61.73) 11,167,413 (61.26)
Separated 25 (0.27) 271 (2.17) 420,140 (2.31)
Divorced 1,070 (11.67) 1,268 (10.13) 2,306,990 (12.65)
Widowed 1,597 (17.42) 1,826 (14.59) 2,870,668 (15.75)
*Ethnicity White 8,845 (96.49) 11,792 (94.54) 17,027,207 (93.40)
Non-white 322 (3.51) 681 (5.46) 1,202,686 (6.60)
*NSSEC Higher Managerial & Professional 3,167 (34.74) 3,982 (32.36) 5,475,646 (30.04)
Lower Managerial 1,249 (13.70) 1,922 (15.57) 2,624,312 (14.40)
Intermediate Occupations 1,059 (11.62) 1,298 (10.52) 2,080,200 (11.41)
Lower supervisory & Technical 846 (9.28) 937 (7.59) 1,457,493(7.99)
Semi routine, routine & others 2,794 (30.65) 4,204 (34.06) 6,592,242(36.16)
*Education NVQ 4 & 5 1,551 (17.06) 2,191 (17.50) 6,965,766 (38.21)
Higher Education, below Degree 1,268 (13.94) 1,459 (11.65) 1,773,428 (9.73)
NVQ3 752 (8.27) 1,173 (9.37) 1,930,079 (10.59)
NVQ2 1,703 (18.73) 2,336 (18.66) 991,560 (5.44)
NVQ1 366 (4.03) 514 (4.11) 1,254,681 (6.88)
Foreign/Other 1,043 (11.47) 353 (2.82) 4,180,862 (22.93)
No Qualification 2,410 (26.50) 4,444 (35.49) 1,133,517 (6.22)
Health Excellent, good 6,330 (72.10) 8,010 (63.97) 11,213,005 (61.51)
Fair 1,729 (19.69) 3,079 (24.59) 4,856,203 (26.64)
Poor 720 (8.20) 1,425 (11.38) 2,160,685 (11.85)
*Smoke Never smoker 3,275 (36.73) 5,777 (46.14) –
Ex-smoker 4,790 (52.25) 4,910 (39.21) –
Current Smoker 1,102 (12.02) 1,802 (14.39) –
*BMI Underweight 69 (0.93) 85 (0.82) –
Normal 1,958 (26.52) 2,761 (26.58) –
Overweight 3,029 (41.03) 4,380 (42.16) –
Obese 2,327 (31.52) 3,163 (30.45) –
Knee Osteoarthritis Yes 1,728 (18.85) – –
No 7,441 (81.15) – –
TOTAL 9,169 (100) 12,521 (100) 18,229,893 (100)

Variables with missing data in HSE & ELSA.

Table 3. ELSA Univariate and Multivariate Logistic Regression Analysis – F test and P values
Variables Univariate Final model
F - test P value F -test P value
*Age 3.01 0.0172 0.43 0.9038
Sex 22.74 <0.0001 20.67 <0.0001
NSSEC 19.54 <0.0001 6.15 0.0001
Health 174.26 <0.0001 54.06 <0.0001
BMI 47.93 <0.0001 24.81 <0.0001
†Illness 234 <0.0001 76.7 <0.0001
Marital 1.98 0.0953
Ethnicity 1.85 0.1741
Education 5,847 <0.001
Smoke 5.85 0.0029
Alcohol 14.14 <0.0001
Fruit & Vegetable consumption 3.94 0.0196

Not significant but included in the final model.

Found to be defined differently in the census and was subsequently dropped from analysis.

Table 4. ELSA Goodness of fit test
Model parameters Archer-Lemenshow GOF values
Age,Sex, NSSEC, Illness, Health, BMI 0.3374
Age, Sex, NSSEC, Health, BMI 0.4861
Age, Sex, Health, BMI 0.1147
Age, Sex, BMI 0.6283
Age, Sex, NSSEC 0.6708
Age, Sex,Health 0.7564
Age, NSSEC 0.3453
Age, Sex, NSSEC, BMI 0.9742
Age,Sex 0.5680
Table 5. HSE Univariate and Multivariate Multinomial Regression Analysis – F test and P values
Variables Univariate Final model
F - test P value F -test P value
Age 2.88 <0.0001 8.87 <0.0001
Sex 41.49 <0.0001 35.73 <0.0001
NSSEC 7.22 <0.0001 1.98 0.0139
Health 35.24 <0.0001 28.75 <0.0001
Marital 13.87 <0.0001 28.89 <0.0001
Ethnicity 3046.85 <0.0001 1236.81 <0.0001
Education 5.7 <0.0001 2.29 0.0016
Smoke 12.54 <0.0001 14.63 <0.0001
Fruit & Vegetable consumption 1.06 0.3843
Table 6. Frequency distribution of variables in Original and synthetic datasets
Variables/Categories Census (%) Hybrid, n (%) Final, n (%) ELSA, n (%) HSE, n (%)
Age 50–59 years 35.09 35.02 34.59 22.49 32.67
60–69 years 31.16 31.16 31.19 37.64 32.09
70–79 years 20.37 20.48 20.75 26.66 22.66
80–89 years 11.16 11.15 11.24 10.96 11.05
≥90 years 2.22 2.19 2.21 2.26 1.53
Sex Female 52.98 52.84 54.07 55.58 53.96
Male 47.02 47.16 45.92 44.42 46.04
NSSEC Higher Managerial & Professional 30.04 30.16 31.39 34.74 32.36
Lower Managerial 14.4 14.29 13.84 13.70 15.57
Intermediate Occupations 11.41 11.49 11.39 11.62 10.52
Lower supervisory & Technical 7.99 8.08 8.17 9.28 7.59
Semi routine, routine & others 36.16 35.98 35.20 30.65 34.06
BMI <18.5 (Underweight) – 0.98 0.98 0.93 0.82
18.5–<25 (Normal) – 26.20 26.39 26.52 26.58
≥24–<30 (Overweight) – 42.20 41.51 41.03 42.16
≥30 (Obese) – 30.62 31.10 31.52 30.45
Knee Osteoarthritis Yes – – 18.69 18.85 –
No – – 81.31 81.15 –
Table 7. Validation parameters (SAE, R2 and SEI) for BMI and OA spatial microsimulations
Variables/Categories BMI simulation OA Simulation
SAE R2 SEI SAE R2 SEI
Age 50–59 years 0.0186 0.9933 0.9918 0.0433 0.9681 0.9569
60–69 years 0.0132 0.9983 0.9980 0.0200 0.9959 0.9951
70–79 years 0.0295 0.9933 0.9931 0.0560 0.9768 0.9753
80–89 years 0.0304 0.9949 0.9949 0.0617 0.9807 0.9800
≥90 years 0.0916 0.9801 0.9797 0.1741 0.9416 0.9289
Sex Male 0.0344 0.9847 0.9792 0.0680 0.9370 0.9327
Female 0.0304 0.9851 0.9847 0.0777 0.9367 0.9064
NSSEC Higher Managerial & Professional 0.0199 0.9978 0.9977 0.0583 0.9908 0.9800
Lower Managerial 0.0374 0.9888 0.9872 0.0818 0.9550 0.9450
Intermediate Occupations 0.0300 0.9956 0.9954 0.0542 0.9867 0.9850
Lower supervisory & Technical 0.0271 0.9940 0.9929 0.0556 0.9788 0.9710
Semi routine, routine & others 0.0124 0.9987 0.9984 0.0353 0.9938 0.9881